Mapping the regional food economy of the North Bay, by function: Sonoma County is the system currently being mapped most deeply and Sonoma Valley the current microscope; Louisville / Kentuckiana is a comparison system used to test and challenge what we see. This is a working RESEARCH geography - no venture geography has been selected. Entities are evidence; verdicts are held to what the evidence supports.
MAP BEFORE INVENTINGLISTEN BEFORE SOLVINGFIND THE SEAM
Updated Sept 14, 2026Map seed r10 · 87 nodesFunction map v1.3Broader geography search paused, not closed (scanner archived, link at foot)VERIFIEDREPORTEDINFERENCEHYPOTHESISUNKNOWNRESEARCH TARGET
Five-minute orientation
each answer is the short version; every section below is the long one
1. How does aggregation / wholesale / distribution actually work?
It is a stack of functions, not one mysterious business: someone buys (or consigns) farm product, pools it, keeps it cold, sells it, trucks it, invoices it, and eats the losses. At industry scale the stack typically appears to live inside one company, on the reading that thin margins only cover it when trucks, warehouse and sales share overhead INFERENCE. Sysco performs it nationally at 18.4% gross and 3.8% operating margin VERIFIED - the scale benchmark that guards against distribution romance. Plain-English primer ↓
2. How does the Sonoma food system work now?
Many tiny, aging farms in a winegrape-dominated economy; modern dairy processing but almost nothing else in the Valley; one farmer/worker-owned aggregator (FEED, ~$5M reported sales) plus one verified rail to the metro (Daylight); inbound broadline from out-of-county depots; local retail access time-rationed to market hours; big public institutional demand; organized rescue. The whole chain ↓
3. Where is real friction vs assumption?
Best-supported friction signals: processing geography (no production-grade shared kitchen appears on the county commissary list reviewed to date for Sonoma Valley), land access for produce growers, and coordination (rescue found ~150K lbs unrecovered while capacity existed). NOT currently supported: "Sonoma lacks aggregation/distribution," "producers can't find demand," "restaurants can't source local." Evidence and counter-evidence ↓
4. What still blocks a conclusion?
FEED's real capacity and economics; Coastline's actual model; broadline's local share; what producers say constrains them; what buyers say local sourcing costs them; whether existing kitchens are full; institutional contracts. UNKNOWN is the research agenda, not a failure. The full list ↓
5. What are we testing next?
The next tests are organized around the key uncertainties: the FEED completion interview, a Coastline conversation, the Oliver's conversation, producer listening (Maggie and Melissa first), buyer interviews with the switching question, the retail field audits, processing/utilization, and institutional procurement records. Every action names the result that would change our thinking. Question → evidence → who answers ↓
The second lens: what is this system optimized for?
The conventional middle appears built to reward scale, price, consistency, route density and buyer convenience, and appears very good at those HYPOTHESIS. The open question is what it under-rewards (producer share, diversified small farms, regional processing, resilience) and what would have to change for the outcomes we want to be the profitable ones. Hypotheses, not conclusions. The lens ↓
Central questions - unchanged, plus one
WHAT NEEDS, IF ANY, ARE NOT BEING MET BY FEED, FARMERS MARKETS, FARM STANDS AND THE EXISTING NETWORK? ("if any" is load-bearing; "no important unmet need" is a valid finding)
WHERE IS SONOMA MISSING PHYSICAL CAPACITY, AND WHERE IS IT MISSING COORDINATION BETWEEN CAPACITY THAT ALREADY EXISTS?
WHAT DOES THE CURRENT FOOD ECONOMY REWARD, AND WHAT WOULD HAVE TO CHANGE FOR IT TO REWARD THE OUTCOMES WE WANT MORE OF?
WHAT DOES FEED ALREADY SOLVE, WHERE DOES IT STOP, AND IS THERE AN IMPORTANT FUNCTION AROUND IT THAT REMAINS UNSERVED? (a valid conclusion: FEED already solves the relevant middle well enough and Almanac has no new operating role there)
Updated 9/14: CURRENT THESIS clarified (shared functions between producer and buyer; supply forward / demand backward) - verdict, null and kill criteria unchanged. Plus FOODSHED FLOW- what this foodshed produces vs where it goes, what it eats vs where that comes from, what "local" actually claims, every touch in the chain, and the both-sides-win economic hypothesis put up to be broken.
The North Bay landscape
geographic orientation for the CURRENT RESEARCH geography · schematic, not a site-selection map
winegrape-dominant mixed farms, winegrape-leaning pasture / dairy / diversified coastal, marine-influenced urban / town plain current research microscope
Schematic orientation map: boundaries, subregions and agricultural shading are simplified and INTERPRETIVE (drawn from the project's qualitative research, not parcel or land-use data). It shows the landscape being studied, not a site. These are not one homogeneous agricultural region: Napa reads winegrape-dominant; Sonoma Valley is wine country inside a more mixed county; Petaluma and the south county carry the pasture/dairy and processing-infrastructure story (FEED, Barn 5400, Clover, Central Milling); the west county and coast are a different landscape and economy.
The research hierarchy
NORTH BAY broader regional context and frame
↓
SONOMA COUNTY the system currently being mapped most deeply
↓
SONOMA VALLEY the current microscope: access, context, live research
Alongside, not inside: LOUISVILLE / KENTUCKIANA - a comparison system and reality check (Maggie), used to test and challenge what we see here. Details below.
Research posture, not a siting decision
THIS IS THE CURRENT WORKING RESEARCH GEOGRAPHY. SONOMA COUNTY AND SONOMA VALLEY ARE THE PRINCIPAL FIELD OF INQUIRY BECAUSE OF ACCESS, CONTEXT AND LIVE RESEARCH. NO VENTURE GEOGRAPHY HAS BEEN SELECTED. THE BROADER GEOGRAPHY SEARCH IS PAUSED AND ARCHIVED, NOT CLOSED.
Current working hypothesis - on trial
thesis clarified 9/14; adversarial structure, null and kill criteria unchanged · the job of this section is still to make it easy to kill
UNPROVEN
What we are currently circling (clarified 9/14)
Share and coordinate the expensive functions between producer and buyer - aggregating supply forward and demand backward - so independent producers have more capacity to produce, diversify, add value and capture more value, while regional purchasing becomes easier and economically credible for buyers.
The functions in question: aggregation · storage · processing · packing · sales · procurement · ordering · invoicing · transportation · route coordination · distribution · inventory · provenance · demand forecasting · buyer coordination · producer coordination. Not all of these belong in one company.
Principle: SHARE THE COMMON FUNCTIONS. PRESERVE PRODUCER INDEPENDENCE AND DIFFERENTIATION. Producers stay independent businesses that compete and differentiate at the product level.
This is a hypothesis, not a direction, and it is NOT simply "build a local food distributor." Physical aggregation and distribution are core to the operating method - the middle physically moves food between producer and buyer - but they are not the entirety of the model, and every touch has to earn its place.
The null - equal standing
CENTRALIZED SCALE MAY STILL BE CHEAPER AND BETTER.
The broadline chain has MORE touches and lower landed cost today I; touch count predicts nothing.
Sysco nets ~3.8c on 18.4c gross at national scale V - the middle is already run thin.
Three middles already coexist in this county (FEED, Coastline, Oliver's procurement) plus direct channels; no strain signal has been observed - though none has yet been looked for directly.
Grasshoppers closed on scale economics, in its own words V; Wave 1: hub math is philanthropy-dependent at scale V.
Operators name land, labor, tenure, housing as binding R - a better middle moves none of them.
The core operating idea (hypothesized model - not an existing flow)
THE MIDDLE AGGREGATES SUPPLY FORWARD →
INDEPENDENT PRODUCERS many small operators, each currently performing its own downstream work
↓
COLLECTION / AGGREGATION consolidate, cool, store, pack, process where it earns its place
↓
REGIONAL MARKET / HUB one sales desk, one order, one invoice, provenance intact
↓
DISTRIBUTION physical delivery on coordinated routes
PRODUCTION SIGNALS what to grow, raise, process, diversify into; where volume and investment may make sense
↓
PRODUCERS clearer demand, less speculative planting
Not a truck-and-warehouse idea: the middle coordinates supply and demand while physically moving food between them. Physical movement is explicit - FARM → COLLECTION → REGIONAL HUB → BUYER - and every touch (cold chain, safety, packing, storage, credit, route density, substitution) must earn its place. The demand-backward half is entirely untested: no buyer has committed a forecast or an order to anyoneU. FEED already performs part of the supply-forward half R - novelty is not claimed.
The thesis in practice? FEED Cooperative - North Bay positive control
On current public evidence, FEED already performs much of this operating model, in this geography, at meaningful regional scale - while preserving farm-level identity:
INDEPENDENT FARMS31 producer-members + 9 worker-members (V, FEED’s own become-a-producer page, live 2026-09-15); 127 listed vendors in the shop directory (V) — ~96 listed vendors are NOT members. The “50+/60+/80 farms” figures are undated marketing copy and are RETIRED as membership numbers.→AVAILABILITYfarms post what they have R→ONE ORDERING SYSTEMV→HARVEST TO ORDERV, verbatim: “no inventory between cycles”; cutoff 8am Mon/Thu, delivery Tue/Fri→AGGREGATION + COLD CHAINV→ONE INVOICE / RELATIONSHIPV/I→PHYSICAL DISTRIBUTIONV→RESTAURANTS · RETAIL · SCHOOLS · CAMPUSES · HOUSEHOLDSV categories
What FEED DEMONSTRABLY DOESprimary source: Lucy O’Dea, FEED, on the record, Arable Ep. 03, Aug 2025: aggregates, holds, distributes and manages sales for North Bay farms · consolidates farm deliveries · runs a shared online ordering platform across participating farms · is a producer AND worker cooperative with patronage split by sales, producers taking a larger share, one member one vote · lets farmers largely set their own prices within a wholesale model · reported ~$5M annual sales (verbatim, as of August 2025 — over a year old, unaudited, and not corroborated by any newer source).
STANDING EVIDENCE RULE (Tyler, 2026-09-16):basic produce aggregation and distribution is NOT an unfilled Sonoma gap. Do not present it as one anywhere on this hub. The live question is where existing middle functions stop serving producer or buyer needs adequately, if anywhere — which functions, categories, buyers, producers or geographies remain insufficiently served, if any.
What public evidence does NOT prove: superior producer net economics · buyer economic superiority · profitability · independence from grants · durable route economics · producer satisfaction · binding demand commitments · long-horizon production planning · scalability. All UNKNOWN - the completion interview's territory.
Two conclusions, held together and not contradictory: (1) FEED may be real-world evidence that much of the model is viable enough to exist and persist R/V functions. (2) FEED may be strong evidence that Almanac should NOT duplicate those functions in the North Bay I. A model can validate a thesis while invalidating a new entrant doing the same thing in the same geography.
Reframed 2026-09-16, after two primary sources landed
The live question, restated
Not “is there a missing middle.” Basic produce aggregation and distribution demonstrably exists here and is retired as a gap. The question is where existing middle functions stop serving producer or buyer needs adequately, if anywhere. That is genuinely open and the honest answer may be “nowhere.”
A leading candidate, and the best single FEED question, but NOT the only one:is buyer demand aggregated BACKWARD into production signals, commitments, expansion or diversification decisions?UNKNOWN FEED requires a Product Availability Template (crop-planning documentation) from every vendor V, so a season-ahead plan exists internally. Whether it reaches producers as something they can act on is unknown. No evidence of contracts, floors or committed-demand instruments U.
Other candidates, none established: fill rate and service recovery (FEED publishes no fill rate, substitution policy or short-credit procedure anywhere U) · categories beyond produce · buyer types not currently served well · producers outside the member and vendor set · geographies at the edge of existing routes · value-added and processing · storage and seasonal extension.
GEOGRAPHIC PRECISION, and do not over-read it. Sonoma VALLEY has no aggregation warehouse and truck fleet of its own; Sonoma COUNTY does. FEED’s physical infrastructure is in PetalumaV. This is a fact about where infrastructure sits, NOT evidence of a Valley service gap. FEED’s published delivery zones explicitly include “Sonoma (outside Petaluma)” at an $8-10 fee V, and two consumer pickup hubs plus two vendor farms sit in the Valley V. Whether a Valley address is a Tue/Fri primary stop or a Wed/Sat outlying stop is UNKNOWN and is a direct question, not an inference.
Land access — a serious candidate constraint, and how to hold it
Three voices, two of them primary and on the record, name land rather than distribution as what stops local production expanding.
Lucy O’Dea, FEED purchasing manager, publicly, verbatim V · Arable Ep. 03, Aug 2025: “there are actually farmers that want to expand, that want to grow, that can and can succeed, that are literally unable to access more land. They can’t certainly can’t buy it, but they can’t even lease it.” And: “by making them bigger, you could employ more people. You could drop the price and make it more available to more people.”
Tyler, unprompted, 2026-09-15 call V · transcript 23:57: land economics push acreage to grapes, the produce farmer who does not own land cannot enter, “and so the distributors are buying from wherever the food is being grown that’s affordable to grow.”
Paul’s ProduceR · Arable Ep. 02: demand is not its constraint; labor and land tenure are.
HOW TO HOLD THIS, and it matters. Three correlated statements are not proof that the problem sits upstream of the middle. Treat land, producer capacity, demand and distribution as potentially INTERDEPENDENT until interviews establish causality. Live readings, none preferred: land is binding and a middle intervention cannot reach it · land and demand bind together, so neither alone is sufficient · committed demand is precisely what would make land accessible or financeable · producers name land because it is the most visible constraint, not the only one. October decides.
The question that settles it, asked identically at every producer stop:what stops you growing more? Not what would help you. Then: if a buyer committed to volume and price for three years, what would change?
Project principle - surfaced by this finding
DO NOT COMPETE WITH FUNCTIONING, MISSION-ALIGNED REGIONAL INFRASTRUCTURE SIMPLY BECAUSE WE CAN IMAGINE ANOTHER VERSION OF IT. PARTNER, STRENGTHEN, EXTEND OR CONNECT FIRST. BUILD ONLY WHERE A REAL SEAM REMAINS. (Applies to FEED, Oliver's, Coastline, processors and every good existing actor: we are looking for missing functions, not replacing organizations.)
The model exists in multiple forms national scan - REPORTED, not independently verified
Asset-light market maker: exited warehouse + truck ownership; rides established distribution partners
The question is no longer whether this kind of middle can exist. It is which functions need to exist in a given region, who already performs them well, and what architecture makes them economic. Survival is not proof of profitability, producer economic improvement, or grant independence - none of these examples is claimed to be any of those.
Supply aggregation vs demand aggregation now one of the most important things to test
Many models aggregate SUPPLY. Far fewer aggregate DEMAND far enough backward to change crop planning, volume, labor, acreage, processing, storage, or investment. The gradient:
Not claimed as the missing seam. It is a sharpened question for FEED, buyers, Maggie and producers: who, anywhere in this system, turns buyer demand into signals producers can act on?
North Bay white space, if any: partner / extend / build adjacent / do not enter CURRENT RESEARCH POSTURES - not venture decisions; all revisable on evidence
Meat slaughter / fabrication without a specific demonstrated gap
The Sonoma question, sharpened: WHAT DOES FEED ALREADY SOLVE WELL? WHERE DOES IT STOP? WHAT, IF ANYTHING, SHOULD EXIST AROUND IT? A valid North Bay conclusion: FEED already solves the relevant middle well enough and Almanac has no new operating role there.
Current verdict (unchanged by the 9/14 rephrase)
INSUFFICIENT EVIDENCE. THE 9/14 CLARIFICATION MOVES THE THESIS TOWARD THE COORDINATION END OF THE RANKING BELOW - THE BEST-SUPPORTED GAP TYPE - BUT NOTHING IS PROVEN: THE 9/12 ADVERSARIAL FINDINGS, THE NULL, AND EVERY KILL CRITERION APPLY TO THIS PHRASING UNCHANGED. THE NEXT DECISIVE EVIDENCE IS FEED'S ECONOMICS AND CAPACITY, PRODUCER-SIDE CONSTRAINTS, BUYER SWITCHING/COMMITMENT BEHAVIOR, AND WHETHER ANY FUNCTION AROUND THE EXISTING MIDDLE REMAINS MATERIALLY UNDERSERVED.
Most likely gap type, ranked on current evidence
B · COORDINATION GAPmoderate R - rescue case, land-matching, buyer-vendor coordination cost
E · PRODUCTION-SIDE GAPmoderate R - land access, tenure, labor, housing named binding by operators
A · NO MATERIAL GAPfully live U - three middles + direct channels coexist; strain not yet sought
C · INFRASTRUCTURE GAPweak-moderate H - Valley kitchen (list-scoped), cold storage (unverified); utilization unknown
D · NEW / ADDITIONAL ECONOMIC-MIDDLE CAPACITYweakest-evidenced H/U - no measured avoidable cost anywhere in the record yet
Likely a combination. This ranking reorders fast once FEED, Coastline, Oliver's and producers answer. Support labels are qualitative; the ordering, not a number, is the claim.
Producer value: the point of sharing functions mechanism chain - hypothesized, untested
PRODUCE→DIVERSIFY→ADD VALUEcreate more value: product, processing, differentiation→CAPTURE VALUEretain more of the economics already created
The aim is not merely "pay farmers more." If shared functions reduce duplicated non-core work or improve access to demand, producers may gain CAPACITY: to produce more, diversify, process, develop value-added products, invest, hire, take on customers. ADD value and CAPTURE value are different; we care about both. Every link in this chain is a hypothesis - no producer has yet told us shared functions would free this capacity.
De-anonymize regional food value layer - economic effect unproven
A tomato becomes "tomatoes." Regional beef becomes commodity beef. Farm, place, variety, practice and story are stripped as food moves through the conventional chain. Wine is the regional counter-example: origin, producer, place, varietal and practice create and preserve value.
Principle: do not let regional food become anonymous as it moves through the system. Conceptually (illustrative): not "TOMATOES" but "SUNRAY FARM · BRANDYWINE TOMATOES · SONOMA VALLEY."
Only matters when it carries real identity, origin, practice and provenance - not storytelling veneer. Possible effects (differentiation, buyer sourcing communication, regenerative visibility, loyalty, producer value capture) are hypotheses; retail claim-vs-provenance work in Foodshed Flow is the evidence base.
The buyer proposition to be tested, never assumed
NOT "pay more because local is good." The test: can regional purchasing become easy enough, reliable enough and ECONOMICALLY CREDIBLE enough that buyers actually change behavior? Economically credible = ideally no increase in total landed cost; realistically no substantial increase where additional value is genuinely created.
Possible buyer value: reliability, service, ordering simplicity, fewer fragmented vendors, provenance and producer identity, practice visibility, regional/regenerative access. Whether buyers value or pay for any of it is exactly what the switching question measures.
The three-party economic test all three must work; the region is the outcome, not the payer
PRODUCER: captures more value, sheds duplicated non-core work, gains capacity to grow/diversify/add value.
BUYER: pays no more (or not substantially more where real value is added), keeps service/reliability/convenience, gains meaningful sourcing value.
MIDDLE OPERATOR: earns enough on the functions that remain necessary to operate durably WITHOUT permanent subsidy.
The REGION (security, resilience, diversity, retained value) is the desired system-level outcome - never a fourth party paying for the model. If one party only wins because another loses, the model is weak.
The case FOR
Producers individually perform expensive sales/delivery/retail functions (the bejkr's 3.5 hr/wk; every direct route) V/I; FEED's consolidation logic demonstrably removes duplication R.
Buyer coordination cost is real testimony (the 15-vendor problem) R; VB+B / Swim Club prove local buying is practiced R.
Oliver's runs county-defined local purchasing at ~30% of COGS inside a growing $180M grocer V self-reported - and its capabilities are internal; smaller buyers cannot access them H.
Manzana's exit creates a dated, named market problem for apple growers (2030) V.
Provenance does not survive the conventional chain ("local" = anything from "our region" to an 8-hour drive) V.
Institutional demand is public, recurring, policy-pushed local V; small-farm matching plausibly unsolved H.
The case AGAINST
FEED, Coastline, Oliver's, markets and broadline already cover the functions; burden of proof is on "something is missing" I. FEED cuts both ways: it may support the thesis at the MODEL level while being strong evidence against an Almanac entry into the same functions in this geographyI.
Operators' own binding constraints are upstream of the middle: land, labor, tenure, housing R.
Supply is small (~250 produce acres, VERIFY), aging (45% of producers 65+), partly leased V/R - volume for a scaled middle may not exist.
Dedicated local middles chronically fail on scale economics (Grasshoppers; Wave 1 hub math) V.
Stated local preference rarely converts at parity, and no Sonoma demand measurement exists I/U.
Cross-haul is often rational; most touches pay for themselves; centralization's core benefits (leverage, redundancy, working capital, forecasting) do not shrink to foodshed scale I.
"Keep food in the foodshed" is a social preference, not by itself a market failure I.
Kill criteria - retire or radically rethink the thesis if:
FEED shows material unused capacity and no category/customer walls
Buyers will not shift meaningful spend at price-and-service parity; provenance does not move orders
Removable cost proves < ~5c per buyer dollar after matching service, or lost scale benefits exceed touch savings
Coastline or Oliver's already gives small buyers adequate local access
Required route density exceeds plausible Sonoma demand
Supply cannot expand because land/labor bind
The model only pencils with standing subsidy, philanthropy or premium pricing
Cross-haul cases prove economically rational on inspection
The partnership criterion: if FEED already performs the core functions well, has capacity to expand, producers value it, and buyers can access it at workable economics, Almanac should NOT recreate those functions - the response is to help FEED expand, connect adjacent infrastructure, or do nothing
What would strengthen it
FEED at or near capacity, with a waitlist or named category walls
Buyers naming concrete dollars they would shift, and what the incumbent fails to do
Producers ranking market access / sales labor in their top two constraints
A measured landed-cost gap in even one category
Oliver's mechanics revealing functions smaller buyers want and cannot get
Manzana growers with no buyer path for 2030
Where we may be fooling ourselves - the leaps, named self-critique
The four layers must not blur: DESIRED OUTCOME (viable farms, value retention, provenance) ≠ OBSERVED PROBLEM (none of "distribution failure" is in evidence) ≠ MECHANISM (aggregation) ≠ VENTURE (build a distributor).
Outcome-to-problem leap: we want value retention; we have not observed a distribution failure - the direct testimony points upstream (land, labor, tenure).
Authority gravity: Michael's $100M framing, Maggie's cost complaint, the Berry lunch - all framing or single-operator testimony, all pulling toward distribution.
Touch-lens aesthetics: short chains feel efficient; the cheaper chain today has more touches.
Perception-gap conflation: the local-claim gap is real; that anyone will PAY to close it is unshown.
Oliver's flattery: 29.6% is local-SUPPLIER COGS, not Sonoma-grown food.
Processing inflation: list-scoped, utilization-blind friction signal, no more.
Rising Tide as attractor: the hypothesis phrasing still echoes a shape that arrived before evidence.
Anecdote inflation: Melissa's "absolutely" is one polite sentence.
Mostly desk, little listening: Tyler has had limited direct conversations (Melissa Bucklin-Good of Oak Hill among them), but the designed interviews - producers, buyers, FEED, Coastline, Oliver's - have not yet happened; most evidence remains secondary, and selection bias stays unknowable until they do.
The 9/14 rephrase itself: "shared functions + demand backward" describes part of what FEED already does (one platform, consolidation) and what a co-op or incumbent could bolt on; the rephrase moves the thesis toward better-evidenced territory but does not create evidence, and it must not be read as progress.
Method discipline (from the RACO deck, method only): UNDERSTAND → DEFINE comes before IDEATE. We are still in the problem space; the hypothesis's "designed to reduce steps" wording is solution-space language that crept in early. RACO's map surfaced a specific missing function (capital orchestration); ours has not yet surfaced an equivalent - and possibly never will. That is a legitimate result.
The system today
the real route choices, with the foodshed boundary visible · solid = verified category or edge; dashed = research target
THE FOODSHED · SONOMA COUNTY (current research geography; Sonoma Valley = the microscope)
SONOMA PRODUCER3,097 farms; 43% are 1-9 ac→DIRECTmarkets · stands · CSA · restaurant back door→CONSUMER / CHEFprovenance by construction; time-rationed
GROCERY SHELVESOliver's · Sonoma Market · WF · Safeway←MIXED CHANNELSdirect + local distributors + DC supply; Oliver's = mixed, verified
RESTAURANT / HOTEL / SCHOOL←BROADLINE + PRODUCE HOUSESthe default inbound machine
↗ OUTBOUND - what leaves
Wine, far beyond the foodshed I
Branded organic dairy (Straus, Clover) region-wide I
Manzana apple products (through the 2029 wind-down) V
FEED product via Daylight to metro foodservice V
Cattle, for out-of-county slaughter I
Petaluma eggs, regionally I
↘ INBOUND - what enters
Broadline full-line supply from out-of-county depots (Sysco: Fremont) V
Chain-retail DC supply (Safeway, WF) V category
Conventional produce via wholesalers; local share RESEARCH TARGET
Most of everything eaten here, by any reasonable reading I - unquantified
Money and touches
follow the money, not just the food · the missing numbers are the point
11.8¢
farm share of the ALL-FOOD dollar, US 2024 V, USDA ERS
~24¢
farm share of the FOOD-AT-HOME dollar, US 2023 V, USDA ERS
18.4 → 3.8¢
Sysco gross → operating margin per sales dollar, FY2025 V - scale benchmark
?
FEED's spread + producer payout. The single most important number we do not have. UNKNOWN
Chain (buyer pays $1.00)
Producer share
The middle takes
What the middle delivers for it
Status
A · Farm direct
~85-95¢ gross; NET unknown after 30-50¢ of the farmer's own retail/sales/delivery labor
none - the farm IS the middle
provenance; price control; capped volume
SCENARIO
B · Farm → FEED → buyer
farmer sets price R; share of buyer $ unknown
FEED spread = ?
aggregation, cold chain, one order/invoice, delivery, patronage
UNKNOWN - decisive
C · Farm → wholesaler → buyer
unknown locally
produce-house gross commonly 20-30% I
risk absorption, cold chain, credit, reliability
I + U
D · Large producer → broadline → buyer
lowest of the chains I
18.4¢ gross → ~3.8¢ kept V at Sysco scale
assortment, density, credit, subs, reps, tech - the full bundle
V benchmark
E · Farm → FEED → Daylight → buyer
compressed by two spreads I
two stacked margins
metro reach local product can't get alone
I
F · Oliver's-style regional retail
wholesale price, one spread when direct
retail gross commonly 25-35% I; Oliver's uses mixed channels (direct + local distributors; FEED a verified produce partner, depth/volume unknown) V
shelf, merchandising, provenance opportunity
I + U
The honest arithmetic — DEMOTED 2026-09-16. The removable slice (duplicated freight, per-farm sales and invoicing, stacked second spreads) was previously put at 5-15¢ of the buyer dollar. SPECULATION — no longer load-bearing That figure is a scenario built on assumptions, not a measurement, and it should not be used to argue the case.
What the public margins DO and DO NOT tell us. Sysco runs a 3.66% operating margin; Chefs’ Warehouse 3.5%, on a 24.2% gross margin it converts LESS efficiently than Sysco does V, FY filings. Read this as a warning about the difficulty and scale economics of distribution: the function is performed for thin operating margins and higher-touch service costs more, not less.Do NOT read it as evidence that no removable inefficiency exists. Thin aggregate margins at national scale say nothing about duplication at the farm-to-buyer edge in one region. Those are different questions and only the three knowable numbers settle the second one: FEED’s spread, Coastline’s economics, and one buyer’s landed-cost decomposition.
The touch test - stressed
A touch that costs 5¢ but saves 10¢ elsewhere is not inefficiency.
Look removable / combinableH
Duplicated farm-truck freight to the same buyers (FEED already removes some)
Per-farm sales, ordering and invoicing against the same buyer
Second spreads added purely for reach (the Daylight hop) - IF direct route density existed
Expensive but earn their keepI
Grading/packing (spec + safety) · DC warehousing (turns, substitution) · dense route delivery · credit/working capital · cold chain · storage across seasonality
Unknown
FEED's internal touch costs · Coastline's · every Sonoma-specific number
What centralization actually solves
The central design constraint for any foodshed-scale idea.
Could plausibly exist at foodshed scaleH
Demand pooling (FEED does it) · short-radius route density in a compact valley · shared cold chain · one order/one invoice · provenance by construction
Depend on scale we cannot recreateI
Procurement leverage · national redundancy and substitution breadth · cheap working capital and credit programs · forecasting on large data · amortized technology · food-safety systems overhead
If the value buyers actually price sits mostly in the right column, the thesis dies on this card.
Foodshed flow
added 9/12 · what we produce vs where it goes; what we eat vs where it comes from; every touch, named and classified · the shorter-chain idea is here to be BROKEN, not sold
Foodshed questions - added to the central set
HOW MUCH OF THE FOOD PRODUCED HERE STAYS HERE - AND HOW MUCH OF THE FOOD CONSUMED HERE COMES FROM HERE?
WHERE ARE WE EXPORTING FOOD WHILE IMPORTING A COMPARABLE PRODUCT BACK IN?
WHICH TOUCHES ADD NECESSARY VALUE, AND WHICH MAY EXIST BECAUSE OF HOW THE SYSTEM HAS HISTORICALLY BEEN ORGANIZED?
CAN A REGIONAL MIDDLE MAKE THE PRODUCER BETTER OFF WITHOUT MAKING THE BUYER WORSE OFF?
WHEN WE SAY "LOCAL," WHAT EXACTLY ARE WE CLAIMING?
The number we are trying to get to RESEARCH GOAL, not a claim
"North Bay residents and businesses consume roughly X million pounds of these foods annually. North Bay farms produce roughly Y million pounds. FEED moves Z million pounds, of which A% comes from the five-county foodshed and B% is sold back into it."
WE DO NOT KNOW THESE NUMBERS YET.
1. How much food does the North Bay consume?
2. How much food does the North Bay produce?
3. How much food does FEED actually move?
4. Where does that food come from, and where does it go?
Public data first. Where it runs out, ask FEED, producers and buyers directly. Rule: DON'T ESTIMATE WHAT AN OPERATOR CAN TELL US - FEED's annual pounds, source and destination geography, category mix, member vs vendor volume, and unused capacity are interview questions, never modeled estimates.
Provenance zones + the six senses of "local" (project standard from now on)
SONOMA VALLEYSONOMA COUNTYNORTH BAYBAY AREA / NORCALCALIFORNIAOUTSIDE CA
Tracked separately for every product: GROWN/RAISED/CAUGHT · PROCESSED · PACKED · AGGREGATED · DISTRIBUTED · SOLD. A product can be local in one sense and not another. Claim taxonomy (not interchangeable): Sonoma Valley grown · Sonoma County grown · North Bay grown · CA grown · raised in Sonoma · processed in Sonoma · packed in Sonoma · distributed by a Sonoma company · made by a Sonoma brand · locally owned · "local" with no definition.
View 1 · What we produce → where it goes
Category
Known first destination
Where value goes next
Share staying in foodshed
Winegrapes (~73% of crop value)
Wineries (in- and out-of-county) V
Finished wine sold far beyond the foodshed I; unsold-grape share verified for named growers
UNKNOWN
Organic milk / dairy
Straus (Rohnert Park) + Clover (Petaluma) - real IN-county processing V
Branded products distributed Bay Area and beyond I
UNKNOWN
Apples
Manzana (Sebastopol) processing - lines end ~Q1 2027; buys Sonoma apples only through 2029 V
Processed apple products ship outward; the 2030 question is datable
UNKNOWN
Eggs / poultry (Petaluma cluster)
Regional egg supply chains I
Bay Area retail + foodservice I
UNKNOWN
Diversified produce (~250 ac claim, VERIFY)
SVCFM, FEED, farm stands, some restaurants V categories
Part rides FEED→Daylight to the metro V
UNKNOWN
Cattle / livestock
Mostly leaves for out-of-county slaughter I; North Bay Butchers = new, very small in-county V
Returns, if at all, as boxed beef via broadline I
UNKNOWN
View 2 · What we eat → where it comes from (retail provenance audit, desk pass)
Oliver's Market possible POSITIVE CONTROL
4 Sonoma County stores; 100% employee-owned V.
Defines local = SONOMA COUNTY, and reports 29.6% of cost of goods sold purchased from local producers (2019; 27.4% in 2015; $85.6M of $182.6M sales) V, company-commissioned studies - self-reported %.
Local categories: produce, meat, bakery, deli, specialty, wine/beer V.
TYLER-REPORTED Melissa Bucklin-Good (Oak Hill) shops there and indicated Oliver's buys meaningfully from local farms - consistent with the record, operationally unverified.
UNKNOWN: produce-specific SONOMA-GROWN share; direct vs FEED vs wholesaler mechanics; small-farm integration; seasonal substitution; shelf-level provenance consistency.
Sonoma Market / Glen Ellen Village Market perception vs provenance case
Owned by NUGGET MARKETS (family-owned, Yolo-County-based) V - locally sited, not locally owned in the Sonoma sense.
Nugget uses "Local and CA-Grown icons"; local = "our region," no mileage, no farms named on the pages reviewed V, company pages.
No evidence of any false claim. The live question: can place-name + presentation create a strong local impression without shelf-verifiable provenance? H - field audit decides. Family perception of "mostly Sonoma produce" = consumer anecdote, not data.
Whole Foods (Sonoma) defined, elastically
Local defined store-by-store ("one community at a time"): state borders or a mile radius, regional foragers decide V, company page; historically ≤ one day's drive V, 2011, dated.
Company claims ~25% of produce from "local farms" under its own varying definition self-reported.
Local items generally integrated through regional DCs I; Sonoma-store specifics U.
Safeway (Sonoma) scale supply
"Local" historically defined as within an EIGHT-HOUR DRIVE of the store V, 2011 statement - dated; current policy U.
DC-supplied; shelf-level provenance visibility U. Studying it for capabilities, not criticism.
Sonoma Valley Fruit Basket Tyler hypothesis - UNTESTED
Identified: Sonoma Valley Fruit Basket, 18474 Sonoma Hwy (Boyes Hot Springs), with The Basket (18460 Sonoma Hwy) and Fruit Basket (24101 Arnold Dr, El Verano); owners Omar and Joewad Homran V, Index-Tribune 2021 + listings.
Public record shows produce + general goods + services (Western Union, U-Haul); no public sourcing claims, farm names, or provenance statements found - in either directionF-negative.
Hypothesis (farm-stand look, conventional wholesale supply): HYPOTHESIS; sourcing UNKNOWN; the in-person field audit below is the instrument. No characterization before then.
Grower-verified provenance by construction (SVCFM roster verified; Oak Hill Red Barn; CSAs).
Time-rationed access remains the structural limit V.
View 3 · Touch map - example chains, every touch classified
Touch classificationESSENTIAL VALUE-ADDNECESSARY BUT EXPENSIVEPOTENTIALLY DUPLICATIVEPOTENTIALLY REMOVABLEUNKNOWN · A TOUCH IS A COST (Michael) - AND NOT EVERY TOUCH IS WASTE. No verdict without economics.
Leafy green, via FEED(farm sets price; ~$5M co-op scale)
GROW + HARVEST + FIELD-PACKfarm · farm owns · ESSENTIAL→COOL + AGGREGATEFEED Petaluma · title model U · ESSENTIAL→SELL + INVOICEFEED platform · ESSENTIAL→DELIVERFEED truck · route economics U · NECESSARY-EXPENSIVE→BUYER RECEIVES
Leafy green, via broadline(the incumbent chain; each touch exists for a reason)
Read honestly: the broadline chain has MORE touches and still wins on landed cost at scale I - touch count alone predicts nothing; the question is cost per function.
Apple, via Manzana(the datable case)
ORCHARDSebastopol growers→PROCESSManzana · juice/sauce · ESSENTIAL · ends ~Q1 2027→BRANDED DISTRIBUTION OUTWARDthrough 2029 wind-down⇢2030: WHO BUYS THESE APPLES?the datable question
Milk, via Straus / Clover(the counter-case: processing DOES exist in-county)
NORTH BAY DAIRIES→HAUL + PROCESS + BOTTLERohnert Park / Petaluma · ESSENTIAL→BRANDED DISTRIBUTIONBay Area + beyond · shares U→RETAIL incl. SONOMA SHELVES
Bread(the geography-of-processing case)
GRAINmostly outside foodshed I→MILLCentral Milling, Petaluma · ESSENTIAL→BAKEthe bejkr: Sonoma home → Petaluma facility · commute = the friction signal→SELLFriday market, 3.5 hrs/week · POTENTIALLY EXPANDABLE, not removable
Beef chain deferred to one line: most animals leave the county for slaughter [I]; North Bay Butchers is the tiny in-county exception [V]; per-touch economics U. Value-added trace deferred - evidence too weak this wave.
View 4 · "Local": claim vs provenance
Retailer
What "local" means
Farms named?
Origin visible at shelf?
Sonoma County provenance?
Aggregator used?
Public sourcing policy?
Oliver's
Sonoma County (defined in its studies) V
Partner/vendor features exist V; per-item U
U - field audit
29.6% of COGS from local producers V self-reported; produce-specific U
Yes - FEED listed as an Oliver's produce partner V, partner page; depth U
Economic studies published V
Sonoma Market (Nugget)
"Our region" - undefined miles V
Not on pages reviewed V
U - field audit
U
U
House-brand "local producers" language only V
Whole Foods
Store-defined; state or mile radius V
Sometimes (forager features) R
Varies U locally
U
Regional DC system I
Corporate program pages V
Safeway
Historically within 8-hour drive V, dated 2011
No I
U
U
DC system V
Current policy U
Fruit Basket
No public definition found F-negative
None found F-negative
U - field audit
U
U
None found
Farmers market / stands
The grower is standing there
By construction
Yes
Yes (roster verified)
None
Market rules
The research question is the reasonable-shopper gap: what a shopper thinks "local" means vs what the chain actually supports. No retailer here is accused of anything; the field audit measures the gap.
View 5 · The economic test (hypothesis on trial, null included)
Core hypothesis - visible, falsifiable, NOT a conclusion
CAN A REGIONAL AGGREGATION / DISTRIBUTION SYSTEM REMOVE ENOUGH TOUCHES AND DUPLICATED WORK THAT THE PRODUCER NETS MORE WHILE THE BUYER'S TOTAL LANDED COST STAYS EQUAL OR FALLS - WITHOUT DEGRADING RELIABILITY, AVAILABILITY, SAFETY, SERVICE, CONSISTENCY, TERMS OR DELIVERY?
Total landed cost - what must be counted mostly unmeasured
Case price is not the comparison. The buyer's real cost:
The stronger question is never "will people pay more to support local farms" - it is: CAN VERIFIED REGIONAL SOURCING BECOME ECONOMICALLY RATIONAL FOR BOTH SIDES?
The null hypothesis - equal standing to be tested, not dodged
Scale purchasing, route density and inventory turns may beat any touch-removal savings: the broadline chain has MORE touches and still wins on landed cost today I.
Sysco's 18.4% gross / 3.8% operating margin says the middle is already run thin at scale V - there may be less removable cost than the touch lens suggests.
Grasshoppers named scale economics as what closed it V; FEED's ~$5M co-op survives R - neither proves the both-sides-win condition exists.
A valid outcome of this audit: the idea does not work.
View 6 · Cross-haul + regional retention
Cross-haul all candidates = RESEARCH TARGETS
Definition: the foodshed produces a product or close substitute, ships some out, while local buyers import a comparable product through another chain. Legitimate reasons exist (seasonality, spec, price, volume, contracts, processing, safety, route economics). None of these is asserted as inefficiency.
SONOMA APPLES→MANZANA PRODUCTS SHIP OUTV anchor, through 2029⇢RETAIL APPLES IMPORTED?shelf audit decides
NORTH BAY ORGANIC MILK→BRANDED, DISTRIBUTED WIDELY⇢CONVENTIONAL MILK IMPORTED?I
PETALUMA EGGS→REGIONAL SUPPLY OUTWARDI⇢NATIONAL BRANDS ON LOCAL SHELVES?audit
SONOMA CATTLE→LEAVE FOR SLAUGHTERI⇢BOXED BEEF RETURNS VIA BROADLINE?I
Regional retention - the $1 lens model built, number NOT estimated
Target: of every $1 of regional food spend, how much reaches the producer, stays with local processors, local labor, local distributors - and how much leaves the county and region.
Required to estimate: producer price-received data (interviews), FEED + Coastline economics, Oliver's category data, retail margin structure, county multipliers.
Oliver's published methodology ($100 spent → ~$186 local impact claim, their study) is a template AND a self-reported number to treat carefully R.
No number is fabricated this wave.
View 7 · Why buyers stay - the switching-cost map
Verified industry practice V, trade literature
PRIME VENDOR AGREEMENTS: volume commitment (often ~80% of purchases) in exchange for negotiated margins/pricing.
Deviated / contract pricing; rebate and incentive programs; credit terms as retention tools.
Plus the operating glue: one order, one truck, one invoice; substitutions; emergency coverage; reps; procurement software; predictable specs; route reliability.
Sonoma application HYPOTHESIS until buyers say so
For price-constrained restaurants, hotels and institutions these relationships may be economically sticky even where operators personally prefer regional sourcing. The issue may not be unwillingness: the incumbent system works, and customers may not demand provenance strongly enough to pay for switching friction.
Never claim specific rebates or contracts at a named Sonoma buyer without evidence.
The switching question (now standard in buyer interviews): what would have to be true for you to move 10 / 20 / 30% of food spend to a verified regional channel - decomposed across price, delivery, minimums, assortment, quality, consistency, seasonality, terms, ordering, invoicing, traceability, provenance, customer demand, rebates/contracts, other. And: what does your current distributor do for you that we would have to match?
View 8 · What would change our mind + the instruments
New instruments added to Testing next: Oliver's produce/buying team · Sonoma Market/Nugget · Fruit Basket field audit · Melissa/Oak Hill follow-up. Existing instruments (FEED, Coastline, producers, buyers, institutions) now carry the foodshed questions too.
Retail field audit instrument (phone-ready)
Per store: store / date / time / department. Sample: 10-15 representative produce categories (leafy greens, tomatoes, berries, apples, citrus, potatoes/onions, squash, herbs, mushrooms, one seasonal) PLUS everything explicitly labeled LOCAL. Comparison, not inventory counting.
Per product: product · variety · price · unit · organic/conventional · farm name shown? · brand · label claim · "local" claim? · local defined? · grown/raised where? · processed where? · packed where? · distributor/shipper if visible · PLU/box info · photo # · notes.
Producer flow audit instrument (for interviews)
Products · acreage · seasonal volume · potential additional volume · destination % by zone (Valley / county / North Bay / Bay Area / CA / out of state) · channel shares (direct, FEED, distributor, retail, restaurant, institution, farmers market, DTC) · price received · packing / storage / selling labor / delivery / freight costs · shrink + rejections · payment timing · minimum-volume problems · unsold product · what would let them sell or grow more.
Most important - RANK the constraints: LAND · TENURE · WATER · LABOR · HOUSING · PROCESSING · STORAGE · PACKING · SALES · BUYER DEMAND · DISTRIBUTION · ROUTE DENSITY · PRICE · CAPITAL · REGULATION · OTHER. Aggregation/distribution is NOT assumed to be the binding problem.
Consumer intercept (small, future; non-leading)
1. "When you see 'local' in a Sonoma grocery store, what geography do you think that means?" 2. "Do you think most fresh produce sold here was grown in Sonoma County?" 3. "What would you need to see to know where your food actually came from?" No willingness-to-pay hypotheticals - behavior beats stated intent.
Melissa / Oak Hill follow-up TYLER-REPORTED anchor
Reported from Tyler's direct conversation: asked where she sells and whether Sonoma could use another market outlet, Melissa said "absolutely." A research signal, not a market finding. Follow-up must pin what "market outlet" means to her: retail storefront? farmers market? wholesale buyer? distributor? year-round outlet? higher volume? better pricing? closer geography? easier logistics? more reliable demand?
Secondary lens: jobs and the wine transition community-economics possibility, not a business case
Documented: wine crop value $626.6M, down 12.6%; 2,700 vineyard acres removed; named growers with 10-75% unsold grapes V. Wine-employment change figures: UNKNOWN. A more diversified food economy MIGHT create different jobs (farming, packing, processing, cold storage, warehousing, driving, sales, production) H - but no claim that Sonoma "needs" these jobs, that vineyard workers can simply transition, or that wine acreage can simply become vegetable acreage (produce is more labor-intensive than vines R; labor and housing constraints persist regardless).
What the big companies already teach the borrowable-capabilities hypothesis - test it, try to break it
Scale operators already solve aggregation, storage, forecasting, inventory turns, ordering platforms, route design, credit, substitution, shrink management, vendor compliance and analytics. The open question: can those CAPABILITIES be applied at foodshed scale with verified provenance and more value retained by producers and the region - or does removing scale remove exactly the thing that makes them affordable? Both answers remain live. The instruments above are how it gets decided.
What we are testing next
Q5 · every action resolves an uncertainty: question → evidence → who answers → what would change our thinking
Tests: is aggregation/distribution already substantially covered? Where does FEED hit economic, capacity or category limits? Evidence: throughput, headroom, margins, title model, turned-away business. Changes our thinking if: FEED has large unused capacity (points to partnership or nothing, per the partnership criterion) OR hard category/economic walls (names the actual seam). Either way, FEED is treated as a positive control to learn from and potentially strengthen, never as a competitor to displace.
2 · Coastline conversation
Tests: how a conventional local wholesaler actually operates; how functionally different it is from FEED. Evidence: customer mix, local share, minimums, routes. Changes our thinking if: Coastline already moves meaningful local product (coordination story), or won't touch it and says why (economics story).
3 · Producer listening method set 9/11
Tests: what actually prevents producers from growing revenue or production. Evidence: their ranked constraints in their own words; cost of selling. Changes our thinking if: constraints cluster on land/labor/tenure (per Paul's) rather than routes - the middle is not the seam.
4 · Buyer interviews Street Social first, relationship-led
Tests: is local procurement materially difficult enough to matter? Evidence: vendor counts, time costs, what they'd shift for. Changes our thinking if: buyers say coordination, not supply, is the binding cost - or say nothing is broken.
5 · Producer-side reality check DEFERRED - no producer session yet held
STATUS CORRECTION 2026-09-16. Maggie did not join the 2026-09-15 working session; that session was Michael and Tyler only. The producer-side pressure test is deferred, not dropped, and no producer has yet tested the thesis. The question set below stands for whenever a producer session happens, and much of it is now folded into the October producer stops.
The ask, unchanged: not “do you like our distribution idea.” Rather: from where you sit as a producer, which shared functions actually matter, which are already solved, and where would solving something genuinely change your economics? Decompose the costs: slaughter and processing access, storage, trucking, sales labor, finding buyers, route density, distributor margins, packaging and compliance, working capital, reliable demand, production planning.
The two questions that matter most: “If something like FEED existed for your products and geography, which parts would actually help and which would not?” And: “Is the harder problem aggregating supply, or getting buyers to make reliable enough commitments that producers can act on them?” And the one added 2026-09-16, asked first and in these words:what stops you growing more? Not what would help you. Then: if a buyer committed to volume and price for three years, what would change?
Scale frame, still unanswered: North Bay demand = X lbs · North Bay production = Y lbs · FEED moves = Z lbs · A% sourced from the five-county foodshed · B% sold back into it. We do not know these numbers. Some are researchable; some must be asked of FEED and producers directly.
6 · Processing / utilization pass
Tests: is missing or remote processing actually preventing products and businesses from existing? Evidence: kitchen utilization, baker's account, North Bay Butchers throughput. Changes our thinking if: existing kitchens sit empty (demand problem, not capacity) or are full with waitlists (real constraint).
7 · Institutional procurement records
Tests: is meaningful demand inaccessible to smaller/regional producers? Evidence: SVUSD contract + local share via board minutes; JCPS bid docs as the Louisville mirror. Changes our thinking if: local share is already high (nothing broken) or blocked by specific bid mechanics (a nameable, fixable seam).
8 · Oliver's produce / buying team new, foodshed wave
Tests: what a regional-first grocery buying system actually looks like operationally. Evidence: produce-specific Sonoma-grown share; direct vs FEED vs wholesaler mechanics; seasonal substitution; whether the 29.6% economics work without subsidy. Changes our thinking if: Oliver's shows the both-sides-win condition already operating (positive control), or reveals it depends on scale/margins a new entrant could not match.
9 · Sonoma Market / Nugget new
Tests: the perception-vs-provenance gap at the Valley's identity grocer. Evidence: shelf audit + how "local" icons are assigned; any named farms. Changes our thinking if: provenance is stronger than the desk record shows (the gap is small), or the local impression rests mostly on place identity.
10 · Fruit Basket field audit new; first field target
Tests: Tyler's untested hypothesis that a farm-stand-appearance retailer runs on conventional wholesale supply. Evidence: shelf labels, boxes/PLUs, shipper marks, staff answers; the phone instrument in Foodshed Flow. Changes our thinking if: it sources meaningfully from Sonoma farms (hypothesis dies), or it demonstrates that farm-stand aesthetics and commodity supply coexist a mile from real farms (the claim-vs-provenance gap in physical form).
11 · Melissa / Oak Hill follow-up TYLER-REPORTED anchor
Tests: what "Sonoma could absolutely use another market outlet" actually means. Evidence: her definition (storefront? wholesale buyer? year-round? volume? price? logistics?), where Oak Hill sells now, what share goes through which channel. Changes our thinking if: "outlet" means demand reliability or price rather than a place - different problem, different instruments.
Standing constraint on all instruments
NO CONTACT with FEED, Catalyst or ownership people without Tyler. October field days read buildings and public faces; designed conversations come after. INSUFFICIENT EVIDENCE remains a preferred conclusion over a forced verdict; "no opportunity" is an acceptable result.
The middle, explained
Q1 · plain English first, then the same functions in two real systems
The vocabulary, without jargon
Farm-direct sales
The farm sells straight to whoever eats or cooks it: market stand, CSA box, restaurant back door. The farm performs every middle function itself - selling, packing, delivering, invoicing, absorbing losses. Highest price per unit, highest labor per unit.
Direct farm delivery
Farm-direct plus a route: the farm runs its own truck to buyers on a schedule. The farm is acting as its own distributor. The bejkr selling 3.5 hours a week at one market is the pure case of a producer whose only channel is his own retail.
Aggregation
Pooling product from many farms into one stream so a buyer orders once instead of fifteen times. A function, not necessarily a company - a co-op, a wholesaler, or a school buying group can each perform it.
Food hub
An organization whose stated purpose is aggregating and marketing local farms' product, usually mission-flavored. May or may not take ownership of the food. FEED is a food hub that became a farmer/worker co-op; Grasshoppers was Louisville's, until scale economics closed it.
Wholesaler
Buys product, owns it, resells it to businesses. Ownership is the tell: a wholesaler takes inventory risk and earns the buy-sell spread. If tomatoes rot, that is the wholesaler's loss.
Produce supplier / produce house
A wholesaler specialized in fresh produce, often adding fresh-cut processing (chopped, portioned). Coastline (Santa Rosa, since 1948) is Sonoma's named example; Horton Fruit plays this role in Louisville.
Distributor
A wholesaler with trucks, routes, order desks and credit terms. In food, "wholesaler" and "distributor" are usually the same business; distribution adds the delivery, ordering and financing layer on top of ownership.
Broker
Sells on commission and never owns or touches the product. No trucks, no warehouse, no shrink risk - paid purely for the sales relationship and market knowledge.
Secondary distributor
Buys from other distributors rather than from farms or manufacturers; fills shorts, small drops, and gaps the big trucks won't serve. Daylight carrying FEED product into Bay Area foodservice makes Daylight a second hop, not a farm buyer.
Broadliner
A distributor carrying the full line - produce to proteins to paper towels - so an operator can one-stop shop. Wins on assortment, price and route density. Sysco, US Foods, PFG, Gordon. Most U.S. restaurant food is generally understood to ride on broadline INFERENCE.
Terminal market
A physical wholesale campus where multiple produce houses sell from stalls and buyers shop nightly. San Francisco has one; Louisville has one (Jennings Lane); Sonoma does not - its equivalent function is a truck from somewhere else.
Procurement / institutional procurement
The buyer-side function: specs, vendor choice, ordering, receiving. "Institutional" adds public rules - bids, contracts, reimbursement-rate ceilings - which is why a school district cannot simply "buy local" because it wants to.
Is the world we are circling one business or five?
Both, and the distinction matters. Each function can exist alone - a broker is sales only, a 3PL warehouse is storage only, a hub can aggregate without owning. At operating scale the stack - aggregation + storage + sales + distribution - typically appears to collapse into one produce/food supply company; the standard reading is that thin margins only pay for trucks, cold rooms and a sales desk when they share overhead INFERENCE from industry structure. Sysco runs the whole stack nationally at 18.4% gross / 3.8% operating margin VERIFIED FY2025 - a scale benchmark, not a universal rule for every regional operator CALIBRATED. The Almanac question is never "should someone build a distributor" - it is which functions are weak or missing in a specific place, for specific actors.
Who actually does what - the function ledger
Function
Farm-direct
Food hub / co-op (FEED-style)
Produce wholesaler (Coastline-style)
Broadliner (Sysco-style)
Broker
Buys from the farm / owns inventory
No middle buyer; consumer buys
FEED's title model (buy vs consign) UNKNOWN - interview question
Yes; takes title at purchase
Yes; title at purchase, mostly from consolidators, rarely the farm itself
Never owns
When ownership changes
At the stand / delivery
UNKNOWN
Farm dock or wholesaler dock
Supplier dock / DC
Never
Inventory & shrink risk
Farm carries all of it
Shared by structure: patronage returns profits, so members absorb outcomes together REPORTED
Wholesaler eats spoilage
Distributor eats it; managed by turns and specs
None
Aggregates multiple producers
No
Yes - the core function (31 producer-members + 9 worker-members; 127 listed vendors) VERIFIED
Yes, from many sources; local farm share RESEARCH TARGET
Yes, at national scale via consolidators
Can, on paper only
Warehouse + cold chain
Farm cooler, if any
Cold storage + reefer trucks VERIFIED
Yes
Yes, DC-scale (Sonoma is served from Fremont and other out-of-county depots)
No
Takes orders / owns the buyer relationship
Farmer does, in person
One ordering platform across member farms REPORTED
FEED invoices buyers INFERENCE from wholesale platform
Yes
Yes
Principal invoices; broker gets commission
Routes trucks / delivers
Farmer drives
Own delivery + the Daylight rail outward VERIFIED
Own fleet INFERENCE; routes UNKNOWN
Dense multi-stop routes - the moat
No
Where the margin comes from
Full retail price, minus the farmer's unpaid time
Wholesale spread; profits return as patronage, producers weighted by sales REPORTED
Buy-sell spread INFERENCE; actual margins UNKNOWN
18.4% gross → 3.8% operating at Sysco scale VERIFIED
Commission (low single digits typical) INFERENCE
What the intermediary is paid to do
-
Give small farms one sales channel and buyers one order; keep farmer economics primary
Absorb risk, keep it cold, show up on time
Make an entire menu arrive on one truck at the lowest workable price
Know who needs what before they do
Blanks are not filled because they seem obvious. FEED's title model, credit terms, and Coastline's routes and margins stay UNKNOWN until someone tells us.
Sonoma: the actual middle
Q1–Q2 · solid arrows = verified; dashed red = research target, not a known flow
How product can move today
FARMe.g. Oak Hill, Paul's→RESTAURANTdirect relationshipREPORTED · VB+B / Swim Club buy from local farms, ranchers, fishermen today
FARM→FEED→DAYLIGHT FOODSUnion City foodservice distributor→BAY AREA BUYERVERIFIED edge - the one proven outbound rail
FARMwhich farms?⇢COASTLINESanta Rosa, since 1948→RESTAURANT / GROCERY / INSTITUTIONlocal sourcing share = RESEARCH TARGET
NATIONAL SUPPLY CHAIN→BROADLINESysco (Fremont), US Foods, PFG; CHEF'STORE Santa Rosa→SONOMA RESTAURANT / HOTEL / INSTITUTIONcategoryroute days, minimums, share
SVUSD DEMAND563,192 meals/yr⇢CONTRACT HOLDER?prime vendor unknown⇢SCHOOL KITCHENSboard minutes = the route to the answer
The five operators that define the middle
FEED Cooperative VERIFIED CORE
Function: aggregation + cold chain + sales + distribution + DTC bins, for 50+ member-language farms within a ~130+ listed-producer network (member vs vendor distinct); one ordering platform instead of thirty farm trucks REPORTED.
Economics: ~$5M reported annual sales operator, Ep. 03; patronage returns profits; farmers largely set their own prices; farmer economics stated priority.
Reach: North Bay buyers + Daylight rail to the metro VERIFIED.
Its own named tension: affordability is hard inside the model - 1–10 acre farms cannot lower prices and stay viable REPORTED.
Coastline Distributors, Inc. MOSTLY UNKNOWN
Not to be confused with Coastline Family Farms (Salinas) - different company.
Function: conventional produce wholesaler-distributor, Santa Rosa, operating since 1948 exists.
Customer mix, sourcing mix, local share, fleet, routes, margins, minimums: all UNKNOWN.
Why it matters: it is the closest thing to a conventional control case for FEED, in the same county.
Daylight Foods VERIFIED EDGE
Union City foodservice produce distributor that carries FEED product (apples, pears, persimmons, mandarins named) into Bay Area restaurants and institutions.
The proof that local product can ride conventional rails outward; depth of the relationship UNKNOWN.
Broadline (Sysco / US Foods / PFG) category
Serves the county from out-of-county depots (Sysco: Fremont); US Foods CHEF'STORE cash-and-carry in Santa Rosa VERIFIED.
Solves availability, price and one-truck convenience for conventional demand extremely well; local share and route economics here UNKNOWN.
Direct farm routes observable
Farm stands, CSAs, Oak Hill's Red Barn (Wed/Sat), SVCFM Friday market, FEED bins - the Valley's dedicated local retail totals a few hours a week plus one farm store.
Every direct route means the farm is performing distribution functions itself; that labor is the invisible cost line.
MANZANA: the standard-setting finding VERIFIED
The county's last apple processor is scheduled to end Sebastopol processing (~Q1 2027) and buy Sonoma apples only through 2029.
Known function + known buyer relationships + known end date = a datable question: what happens to those growers' volume in 2030? This is the evidence standard every claimed gap has to meet.
Critical question: how is FEED operationally different from a conventional wholesaler like Coastline?
Where they appear to OVERLAP (operating function)
Both aggregate many suppliers into one order, one delivery, one invoice for business buyers.
Both own cold storage and trucks; both run sales desks and buyer relationships.
Both sell to restaurants, retail and institutions.
On the desk record, the operating function looks substantially similarINFERENCE - which is precisely why "Sonoma lacks a distributor" fails as a claim.
Where they appear to DIFFER
Governance/ownership (fact, not function): FEED is farmer + worker owned, one member one vote, profits return as patronage VERIFIED. Being a co-op does not by itself change what the trucks do.
Price formation: FEED farmers largely set their own prices REPORTED; a conventional house buys at market and sells at spread INFERENCE.
Sourcing pool: FEED = defined North Bay membership; Coastline's mix UNKNOWN.
Whose risk: patronage socializes outcomes among members; a wholesaler's shareholders keep spread and eat shrink.
Magnitude of real operational difference (routes, minimums, service level, category depth): UNKNOWN → this is what the FEED + Coastline conversations are for.
Business models, side by side
Q1 · V = verified · R = reported · I = inference · U = unknown - blanks are not guessed
Model / operator
Geography
Aggregates multiple farms
Buys / owns inventory
Stores
Cold chain
Sells
Delivers
Takes shrink risk
Typical buyer
Local / regional priority
Farm direct
on-farm / market
No
Farm owns until sale
Farm cooler I
Varies by farm
Farmer sells
Farmer drives
Farm, fully
Consumers, chefs
Total - it IS the product
FEED Cooperative
North Bay (Petaluma)
Yes; 31 producer-members + 9 worker-members; 127 listed vendors V
Title model U
Yes V
Cold storage + reefers V
Wholesale platform + bins V
Own trucks + Daylight rail V
Members via patronage R
Restaurants, retail, schools, access programs V
Total - stated purpose
Coastline Distributors
Santa Rosa, since 1948
Multiple sources I
Yes I
Yes I
Yes I
Sales desk I
Own fleet I
Yes I
U
U - the interesting cell
Daylight Foods
Union City → Bay Area
Carries multi-farm product via FEED V
Yes I
Yes I
Yes I
Foodservice accounts V
Yes V
Yes I
Metro restaurants, institutions
Carries local lines among others V
Sysco / broadline
National; Fremont depot serves Sonoma
Via national consolidators
Yes V
DC scale V
Full V
Reps + catalog + contracts V
Dense routes V
Yes, managed by turns
Restaurants, hotels, institutions
Optional program, not the model
What Chefs Want (Creation Gardens)
Louisville HQ; 11 metro hubs
Via Local Food Connection V
Yes I
Yes; 1904 cold-storage roots V
Yes V
Chef-focused catalog V
Own routes V
Yes I
Independent restaurants, chefs
A program inside a conventional model V
Grasshoppers Distribution †2013
Louisville
Yes - farms + artisans V
Yes I
Yes I
Yes I
Wholesale + 1,200 subscribers V
Yes V
Yes
Households, restaurants
Total - and it closed on scale economics V
Marksbury Farm (meat)
Lancaster KY, ~70 mi out
Partner farms' livestock V
Yes, buys animals I
Yes
Yes
Brand: WF, Kroger, chefs, online V
U
Yes
Grocery, restaurants, consumers
Total - KY-sourced brand
Louisville / Kentuckiana: the comparison system
Q1 · Q4 of the brief · first-pass desk map, built to be corrected by Maggie - not a build-location evaluation
Produce & foodservice flows
KY / S. INDIANA FARM→LOCAL FOOD CONNECTIONfood hub, operating INSIDE What Chefs Want→LOUISVILLE RESTAURANT / SCHOOL / RETAILprogram existsfarm count, terms
NATIONAL + REGIONAL SUPPLY→WHAT CHEFS WANT / CREATION GARDENSindependent; Louisville HQ; produce–meat–seafood–dairy→INDEPENDENT RESTAURANTS11 metro marketsVERIFIED
NATIONAL SUPPLY→BROADLINESysco Louisville depot IN market; USF / PFG / GFS regional→RESTAURANTS / HOTELS / INSTITUTIONSSysco depotothers' geometry
GROWER / SHIPPER⇢LOUISVILLE PRODUCE TERMINALJennings Ln - tenants, volumes?⇢SMALL GROCERS / RESTAURANTSstructure exists; flows = RESEARCH TARGET
FARM⇢WHO?JCPS prime vendor unknown⇢JCPS~96K students; Farm to School program existsbid documents are public = answerable
The cattle chain - Maggie's world
KY COW-CALF862K beef cows; 31,060 producers; largest herd east of the Mississippi→STOCKER / BACKGROUND⇢OUT-OF-STATE FEEDLOTPlains⇢NATIONAL PACKER→BOXED BEEF RETURNS VIA BROADLINEINFERENCE - industry-standard flow; VERIFY WITH MAGGIE
PRODUCERenrolls cattle→BEEF SOLUTIONS LLCowned by KY Cattlemen's Assn; in-state processing→KROGER 250+ STORES / YUM CENTERVERIFIEDproducer economics
FOXHOLLOW FARMclosed herd, Crestwood→OUTSIDE USDA PROCESSORwhich one?→OWN MARKET / ONLINE / BEEF SHARES / LOUISVILLE MENUSVmenus per own posts
SMALL PRODUCERwants local customers⇢SLAUGHTER SLOT?wait times, minimums = the national small-plant constraint; KY specifics unknown⇢?MAGGIE QUESTION, not an assumption
What the Louisville record already teaches
Grasshoppers Distribution, 2007–2013 VERIFIED
Louisville's dedicated local-food distributor (farms + artisans + a 1,200-subscriber grocery program) closed in Dec 2013. Their own words: "The economics of the local food distribution business requires scale that we have not been able to achieve."
A useful comparison case, not a verdict. The question it sets up: why did Grasshoppers fail on scale economics while FEED has survived? Candidate explanations (market, model, management, geography, timing) remain UNKNOWN until researched and listened to.
West Louisville FoodPort, canceled 2016 VERIFIED
Seed Capital Kentucky's $53M, 24-acre everything-hub (grow, process, sell, educate; 200 jobs) died in financing when anchor tenant FarmedHere (vertical farm) withdrew during final design. $2.7M of design spend lost.
Infrastructure-first, anchor-dependent, built before committed operators. A caution for any build-it-and-they-will-come instinct - including ours.
The absorption INFERENCE, flag for Maggie
Louisville's surviving food-hub FUNCTION appears to live inside a conventional independent distributor: Local Food Connection (founded as a stand-alone hub, Cincinnati lineage) now operates as What Chefs Want's local-sourcing platform with farm-level traceability.
Three fates for the local middle in one city: die alone (Grasshoppers), never get built (FoodPort), or get hosted by scale (LFC inside WCW). Sonoma chose a fourth: co-op survival at ~$5M.
Also on the access side: Dare to Care (regional food bank), New Roots Fresh Stop Markets (pooled-share pop-up fresh food), Black Soil KY (connects Black Kentucky farmers to buyers), Kentucky Proud (state marketing label) - all exist, depth unquantified. Dairy distribution: UNKNOWN.
Questions for Maggie
she joins Tuesday's call (9/15) as a producer-side resource · we are not asking her to validate a thesis; we are handing her a draft concrete enough to correct · the working-session frame is in Next tests, card 5
The cost question, decomposed
When you say getting cattle/product to customers is one of the biggest costs - which of these is it, in order?
Who actually moves smaller Kentucky producers' food into Louisville - and is Local Food Connection inside What Chefs Want the real answer, or a marketing layer?
Who already solves parts of the cattle problem (Marksbury? Beef Solutions? custom-exempt plants?) and where do they stop?
Where do small farms still fall out of the system entirely?
What do restaurants actually buy through - and what share is WCW vs Sysco vs anyone local?
What do schools / hospitals / universities actually buy through? Who holds the JCPS produce contract?
Is the Produce Terminal on Jennings Lane still a living market or a shell?
Which companies or people matter here that public research will miss?
Where is our diagram wrong?
Sonoma vs Louisville: what is structurally different?
Q6 of the brief · same functions, very different systems - solutions do not transfer by default
Dimension
Sonoma County / North Bay
Louisville / Kentuckiana
Producer scale
3,097 farms; 43% are 1–9 acres; 45% of producers 65+ V
KY statewide: 69,400 farms, 31,060 cattle producers V; metro-county farm profile U
Product mix
Winegrapes ~73% of crop value (contracting); organic dairy strong; small diversified produce/meat V
Cattle #1 (~$1.5B receipts); row crops; diversified produce is the small slice V
The signature animal/crop problem
Land access for produce expansion (~60K grape acres vs ~250 produce acres) R, verify
Herd is huge, in-state finishing/packing footprint small; calves leave the state I, verify
Land cost
Very high; premium wine-country pricing I
Materially lower I; figures U
Labor + housing pressure
Named PRIMARY constraints by operators R, Ep. 02
UNKNOWN - ask Maggie
Processing capacity
Dairy modern (Straus, Clover); meat new but tiny (North Bay Butchers); no production kitchen on the county list reviewed to date for the Valley V, list-scoped
Marksbury (meat, 70 mi), Beef Solutions (in-state), WCW fresh-cut; overall census U
Cold storage
No third-party food cold storage found in-county I from absence; verify
Cold-storage heritage since 1904 (Grocers Ice → WCW); current 3PL landscape U
Route / demand density
Small towns + tourism corridor; metro demand is 45+ min away
Louisville metro ~1.3M I wrapped around the farms' market
Distributor competition
No in-county broadline depot; FEED + Coastline + inbound trucks
In-market Sysco depot + WCW HQ + terminal market structure V
Institutional demand
SVUSD 563,192 meals/yr; universal meals funding V
JCPS ~96K students - an order of magnitude larger V category
Farm-direct / market culture
Dense: markets, stands, CSAs, farm stores V
Present (markets, New Roots) but thinner per capita I
Cooperative infrastructure
FEED alive at ~$5M reported R
Grasshoppers dead (2013); hub function absorbed by WCW V/I
Buyer expectations
Tourism-premium menus; local identity has retail value
Value-driven metro market I; chef segment served by WCW
Seasonality
Mild coastal; nearly year-round production
Continental; hard seasonal gap
Regulatory environment
CA: strict, expensive, but universal-meals + farm-to-school money flows V
KY: lighter regulatory load I; program landscape U
State ag development funds (tobacco settlement lineage) I, verify
Are the same functions actually missing in both places?
On current evidence: no - and that is the finding. Sonoma's sharpest verified signals are processing GEOGRAPHY (capacity exists, but not in the Valley), LAND ACCESS for produce growers, and COORDINATION between capacity that already exists. Louisville's sharpest signals (pending Maggie) look like MEAT PROCESSING ACCESS at small-producer terms and the SCALE ECONOMICS of a dedicated local middle - the reason Grasshoppers itself named when it closed. Similar complaints ("getting product to customers is expensive") appear to come from different underlying systemsINFERENCE, held. A solution designed for one does not transfer to the other by default - which is exactly why the method (map functions, then listen) is the portable part.
The whole Sonoma system
Q2 · function first - click any function for players, economics, friction and the held verdict (source: function map v1.3)
LAND / INPUTS→PRODUCTION→HARVEST / PACK→PROCESSING / VALUE-ADD→STORAGE→AGGREGATION→DISTRIBUTION→WHOLESALE / PROCUREMENT→GROCERY · RESTAURANT · HOTEL · SCHOOL→CONSUMER→RESCUE / WASTE
Sonoma County is the system; Sonoma Valley is the microscope. FEED, Oak Hill, Coastline, SVUSD, Sonoma Family Meal and the rest appear below as evidence of the function they perform - the function comes first.
The landscape we are mapping (seed r10)
Work in progress. Categories show the system we intend to understand. Named entities are selected evidence/examples, not the complete population. This is not a census of the regional food economy; a category count appears only where a defensible census exists.
3,097 farms countywide [census]486,444 residents · $104,674 median income~8,000 Valley food-insecure563,192 SVUSD meals/yr
Demand
Restaurants / foodservice LANDSCAPE NOT YET QUANTIFIED
Farmers markets and farm stores are structural demand/retail channels here, not afterthoughts: SVCFM (Fri 9-12:30, roster verified), the Tuesday seasonal market (May-Sept), Santa Rosa Original CFM (countywide), Oak Hill's Red Barn Store all sit in this category's examples. Research still needed: full store landscape, local-share by store, provenance at shelf.
Named entities below (Vella, Spread Kitchen, North Bay Butchers, Straus, Clover, Central Milling, Manzana) are examples inside these functions - not a processing census. Utilization of existing capacity = unknown.
Selected mapped examples
The middle / movement
Aggregation · distribution · storage RESEARCH INCOMPLETE
FEED, Coastline, Daylight, broadline and the storage question sit here - the current center of the inquiry. Throughput, capacity and economics = the interview agenda.
Network + capital (context) programs and funders around the operating system
Selected mapped examples
Click any example for its role, evidence status, verified links and unknowns. Solid connections are VERIFIED; dashed items are research targets. Consumers appear as the demand baselines above, not as dots. The purpose of this view: what parts of the system exist, what we have actually mapped, and what is still blank.
F1 · PRODUCTION verdict: INSUFFICIENT EVIDENCE
Players: 3,097 farms; 43% 1–9 acres; 45% of producers 65+ [V]. Working set: 11 verified SVCFM vendor farms, FEED's 50+ member-language farms, Oak Hill, Paul's Produce, True Grass.
Economics/capacity: grapes ~73% of value and contracting (2,700 vineyard acres out; named growers 10–75% unsold) [V]. Acreage is not scarce; land ACCESS (price, tenure, water) is the live question [calibrated]. Friction: succession, labor; each small farm performs its own downstream functions. Counter-evidence: Paul's Produce reports demand and room to grow - labor and tenure are its constraints [R]. Verdict: production may hold constraints (land access, labor, succession) but is not currently the lead inquiry.
F2 · PROCESSING / VALUE-ADD verdict: STRONGEST CURRENT FRICTION SIGNAL - no more than that
Players: dairy modern and real (Straus Rohnert Park, Clover Petaluma) [V]; meat new but "very small" (North Bay Butchers USDA, Petaluma) [V]; grain (Central Milling, Petaluma) [V]; 25 county-listed commissary kitchens clustered Santa Rosa/Petaluma/Cotati; Sonoma Family Meal's 3,100 sf at $25/hr [V]; Vella closed [V].
The supported pattern:no production-grade shared kitchen appears on the Sonoma County rental/commissary list reviewed to date for Sonoma Valley [V, scoped to that list - absence from one list is not exhaustive proof of absence]. Valley producers travel to Petaluma/Santa Rosa for production space (the bejkr: Sonoma home → Petaluma facility → Friday market) [V/R]. Held carefully: friction signal ≠ business opportunity. Needed: utilization of the existing 25 (are they full?), the baker's own account, producer listening.
F3 · AGGREGATION verdict: INSUFFICIENT EVIDENCE - duplication risk HIGH
FEED is the existing solution (50+ member-language farms, ~130+ listed producers; ~$5M reported sales; patronage; farmers set prices) [V/R]. Rescue-side aggregation exists separately (Extra Food, REFB) [V]. Capacity headroom is UNKNOWN and is the hinge of the central question. FEED's own named tension: affordability is difficult inside the model - farmer viability and consumer affordability are different problems [R]. Wave 1 stands: hub math is brutal and philanthropy-dependent at scale [V].
Capacity lives INSIDE operators (FEED, REFB's 5,000 sf charitable freezer, processors) [V]. First pass found NO third-party food cold-storage warehouse in-county [I from absence - verify before use]. Hypothesis worth testing: storage as a hidden rationing mechanism for small producers without their own cold chain [H].
F5 · DISTRIBUTION verdict: INSUFFICIENT EVIDENCE - the 3.8% warning stands guard
Inbound: broadline from out-of-county depots (Sysco: Fremont) + CHEF'STORE Santa Rosa [V]; effectively unlimited capacity. Outbound for local product: FEED + the verified Daylight rail; thin. Benchmark: Sysco FY2025 18.4% GM / 3.8% OM - scale benchmark, not a universal rule [V, calibrated]. The unresolved fork: does local product face a route-to-buyer constraint at meaningful volume, or a demand constraint? Producer voices push AGAINST distribution-as-universal-bottleneck (Paul's: labor/tenure; FEED staff: land access; FEED already does much of the consolidation job). Do not force every problem toward distribution.
F6 · RETAIL verdict: Rising Tide's front-door territory - UNPROVEN either direction
Conventional capacity ample (Safeway, Whole Foods, Sonoma Market, Glen Ellen Village Market) [V]; DIRECT local access is time-rationed to a few market hours + one farm store [V→I]. Convenience wins everyday trips [I]. The decisive evidence is measurable: Sonoma Market's local assortment share, then demand behavior - not more concept.
Landscape unquantified (restaurant count, price bands, hotel purchasing structures = blanks). Buyer friction [R]: the 15-vendor problem - local sourcing costs coordination time; distributor dependence is rational. Counter-example: VB+B / Valley Swim Club buy from local farms, ranchers, fishermen today [R]. Hypothesis: buyers may want MORE of what FEED partially is (one order, one delivery, one invoice) - an interview question, not a finding.
F8 · INSTITUTIONAL DEMAND verdict: most policy-supported demand pool; POSSIBLE matching problem
SVUSD: 563,192 meals/yr, free to all students, CFCK member since 2015 [V]; Vintage House 400 meals/wk [V]. Universal Meals + permanent CA Farm to School money push demand local [V]. Friction: bid rules, volume/consistency specs vs 1–9 acre farms [I]. Evidence needed before calling the matching problem real: SVUSD's actual contract and local share (board minutes route identified).
F9 · CONSUMER DEMAND verdict: segmentation is the missing analysis
486,444 residents; $104,674 median income; ~8,000 food-insecure Valley residents; heavy visitor overlay [V]. A premium-income county with a real access population and a large low-wage hospitality workforce inside it [V→I]. Missing: which households would change behavior for everyday regional food, at what price [U].
F10 · RESCUE / WASTE verdict: a learning window, not a venture thesis
Organized and recently professionalized: Extra Food (20,000 lbs in month one), REFB (~82,000 people/yr; buys 10% of its supply wholesale), Farm to Pantry, Sonoma Family Meal [V]. What it reveals: the commercial system overshoots at the SHELF, not mid-chain [I, held]. The coordination case [R, Ep. 03]: ~150,000 lbs recoverable food NOT recovered while recovery moved ~20,000 lbs/month; 15+ nonprofits working separately; collaboration raised both supply and access points. Capacity existed; connection didn't.
CROSS-CUTTING · COORDINATION / BROKERAGE the standing question on every function
Recurring pattern: resources exist without connecting - landowner↔farmer, surplus↔access org, producer↔buyer, nonprofit↔nonprofit. FEED performs coordination commercially; the access system performs it charitably. Standing question: where is Sonoma missing PHYSICAL CAPACITY, and where is it merely missing COORDINATION between capacity that already exists? Land-to-food research targets: the 5-acre winery offer; Norman Gilroy's reported 2.5 acres / 50,000 lb goal [R - identify and verify].
CROSS-CUTTING · CONSTRAINT LAYER seeded from operator testimony, Ep. 02–03
Tracked across every function: LABOR · HOUSING · LAND TENURE · WORKFORCE SKILLS · IMMIGRATION & SAFETY · WATER · CAPITAL · REGULATION · SEASONALITY. Live evidence [R]: labor scarcity despite demand (Paul's); housing as PRIMARY constraint (Bucklin); leasing limits investment; produce is far more labor-intensive than vines (conversion is not a land question alone); philanthropy cannot substitute for operating economics; a shared worker-support service logged as a candidate shared worker-support function [R/H].
Per-actor constraints - the anti-collapse synthesis: small produce farmer: land, labor, tenure, sales logistics · buyer: convenience + vendor coordination · low-income household: affordability + access · rescue: coordination + transport · nonprofits: fragmentation · FEED: aggregation largely addressed; affordability still hard.
Signals vs assumptions
Q3 · counter-evidence gets equal billing - that is the method
The standing rule
DIFFERENT ACTORS FACE DIFFERENT CONSTRAINTS. DO NOT COLLAPSE THEM INTO ONE BOTTLENECK OR ONE SOLUTION.
"Sonoma lacks aggregation / distribution" NOT SUPPORTED
FEED already performs meaningful aggregation, sales, cold chain and distribution for a 50+ member / ~130+ listed-producer network at ~$5M reported sales, plus a verified rail into the metro (Daylight).
What IS open Capacity headroom, economics, and the categories/customers FEED cannot economically serve - the completion interview, not the assumption.
"Producers can't find demand" NOT UNIVERSAL
Paul's Produce reports demand and the ability to grow materially - its binding constraints are LABOR and SECURE LAND TENURE REPORTED.
Counter-counter One strong producer is not the population; producer listening decides how general this is.
"Restaurants cannot source local" NOT UNIVERSAL
Valley Bar + Bottle and Valley Swim Club buy from local farms, ranchers and fishermen today REPORTED.
What IS live The 15-vendor coordination cost is real buyer testimony; whether it suppresses local share at scale is the interview question.
Processing geography STRONGEST FRICTION SIGNAL
No production-grade shared kitchen appears on the county rental/commissary list reviewed to date for Sonoma Valley [V, list-scoped]; producers commute to Petaluma/Santa Rosa; meat capacity new but tiny.
Held FRICTION SIGNAL ≠ BUSINESS OPPORTUNITY. Utilization of the existing 25 kitchens is the missing fact.
Affordability vs market access DIFFERENT PROBLEMS
FEED's own staff: affordable food is difficult inside the model - 1–10 acre farms cannot lower prices and stay viable. Better aggregation does not automatically feed anyone.
Implication Producer market access and household food access get separate analyses, separate instruments, separate verdicts.
Rescue reveals coordination, not capacity failure REPORTED
~150K lbs went unrecovered while organizations, food, and transport all existed; collaboration raised recovery AND access points.
Held This does not prove commercial distribution failure - it proves connection can lag capacity.
Cross-system echo: the Grasshoppers epitaph VERIFIED
"The economics of the local food distribution business requires scale that we have not been able to achieve." - Louisville's dedicated local distributor, closing, 2013.
Why it's here The scale question may not be a Sonoma quirk - it recurs across systems HYPOTHESIS. FEED surviving at ~$5M is data; Grasshoppers closing is data; why they diverged is UNKNOWN.
The Manzana standard VERIFIED
Known function + known buyer relationships + known end date (~Q1 2027 lines end; Sonoma apples bought only through 2029) = a DATABLE question.
The bar Every claimed gap must eventually look like this: what function, what flow, what economics, what capacity, what unresolved problem.
The unknowns blocking a conclusion
Q4 · UNKNOWN is the research agenda, not a failure
FEED
Capacity headroom + throughput
Margin / fee structure; producer payout share
Title model (buy vs consign)
Producer + buyer satisfaction
Categories it cannot serve; customers and producers it cannot ECONOMICALLY serve
Which school districts, what volume; LFPA/CalFood share of revenue
Coastline
Customer + sourcing mix; local share
Fleet, route structure
Margins, minimums
Overlap with FEED, real not apparent
Why buyers choose it
Broadline
Who actually serves Sonoma, from where
Route days, minimums
Share of restaurant / hotel / institutional spend
Why buyers stay
Producers
Actual cost of selling + delivering (sales labor, shrink, packing, storage)
The biggest constraint to growing revenue: buyers vs land vs labor vs processing vs logistics
Produce acreage (~250 ac claim) vs the crop report verify
Buyers
Supplier count, ordering + receiving burden
Delivery windows, payment terms, minimums
Local sourcing friction: price, availability, consistency, specs
What would make them shift share
Processing
Utilization of the 25 commissaries (full or empty?)
Rates, equipment, access terms
North Bay Butchers throughput + waitlist
Actual unmet need vs geography inconvenience
Institutions
SVUSD contract holder, bid rules, specs, local share, service days
Hospital + senior procurement
Reimbursement ceilings vs local price points
Storage
Verify the no-3PL-cold-storage finding
Rates and vacancy if any exists
Whether storage rations small producers
Louisville (pre-Maggie)
LFC-inside-WCW: farm count, terms, whether it's substance or shelf talker
JCPS prime vendor + local share
Produce Terminal: alive or shell
KY slaughter wait times + minimums
Why Grasshoppers died while FEED lived
Dairy distribution; Horton's mix
What is the current system optimized for?
the second strategic lens · a research lens, not permission to invent · every entry is a HYPOTHESIS TO TEST
Michael's early instinct - go the direction others aren't - is useful only when it becomes a question: why did the current food system evolve this way, what does it reward, and are those the outcomes we actually want? Not contrarianism; diagnosis.
What the middle appears to reward hypotheses
Scale - overhead only pays for itself across volume (Sysco's 3.8% OM; Grasshoppers' epitaph)
Price + consistency - buyers order what arrives identical every time
Route density - more drops per truck-mile beats everything local about a product
Utilization + inventory turns - cold rooms and trucks must never sit still
Shelf life + standardized specs - what survives the system defines what gets grown
Buyer convenience - one order, one truck, one invoice (the 15-vendor problem in reverse)
Producer share of value - and the sales labor small farms perform unpaid
Smaller + diversified farms - diversity cuts volume-per-crop, the thing the system prices
Regenerative practice, soil + ecological outcomes - invisible to a spec sheet
Regional processing + shorter chains - the middle appears to have consolidated away from places like the Valley I
Seasonal diversity - menus built on availability, not availability built on menus
Worker stability, regional resilience, community economic value
Food access + producer independence - the two ends the middle isn't paid to care about
The question this lens exists to hold open
IF WE STARTED WITH THE OUTCOMES WE ACTUALLY WANT FOR A REGIONAL FOOD ECONOMY, WOULD WE DESIGN THE MIDDLE OF THE SYSTEM THE SAME WAY?
What would have to change - categories of possible system change, not proposals
Demand side
Procurement rules + local preference
Contracts + offtake commitments
Buyer coordination (shared ordering)
The middle
Aggregation + wholesale terms
Distribution routes + shared logistics
Processing capacity + geography
Shared infrastructure (kitchens, cold chain)
Production side
Land access + tenure
Producer coordination
Workforce + housing
Enablers
Financing + incentives
Policy
Information + transparency (who buys what, at what price)
The point is diagnostic: what behaviors does the current economy reward, what practices does it make economically difficult, and what would have to change for practices that address the problems we care about to be rewarded. Do not assume the answer is "build a distributor."
Project evolution
how the current inquiry came from the earlier one · historical, not current
Read this first
EVERYTHING IN THIS SECTION IS ARCHIVED PROJECT HISTORY. IT DESCRIBES WHAT ALMANAC WAS BEFORE THE INQUIRY MOVED TO THE MIDDLE OF THE FOOD SYSTEM. NONE OF IT IS THE CURRENT THESIS.The active operating map is everything above this section. The full pre-pivot hub is preserved unrewritten and linked at the foot of this section.
A · EARLIER EXPLORATION What we were exploring
Place-based hospitality as the lead venture idea
Farm stay / stay-and-eat formats: a property where lodging, food and working agriculture share one operation.
Restaurant and food-business concepts attached to land, and the relationship between a hospitality P&L and an agricultural one.
Hospitality as the funder of agriculture: whether a stay could carry a farm that could not carry itself.
Committed demand from an owned kitchen as a way to give producers a reliable buyer. this idea survived the pivot and is now part of the current inquiry
Land plus operating business questions: buy, lease, or partner on someone else’s ground.
A stronger sense-of-place thesis: that attachment to a particular landscape is an economic asset, not decoration.
Strengthening communities through food systems as the outcome the venture was for.
How to read this now
These ideas were not wrong and were not discarded. The honest framing: earlier formats explored; currently de-emphasized while the regional food-system and middle thesis is investigated.
Hospitality remains live in the current map, in a different role. Hotels, resorts and restaurants are present right now as buyer and demand nodes in the active system map, and three of the October fieldwork stops are buyers of exactly this kind: Valley Bar + Bottle and Valley Swim Club, Street Social, and MacArthur Place hospitality procurement.
What changed is the role, not the relevance. A hospitality property is no longer the lead venture hypothesis. Hospitality as a source of demand, provenance and price tolerance is still part of what the current thesis has to explain.
B · ARCHIVED The geography work
The exercise, as originally agreed: each of Tyler and Michael independently returned with four to five places genuinely worth exploring for Almanac, before doing comparative research. Both halves were completed and were deliberately shown side by side, not merged, not ranked, not scored. Last updated in the archived hub September 1, 2026.
Tyler’s five archived
North Bay / Sonoma, Napa & Marin. Personal centre of gravity, plus the North Bay’s relationship with San Francisco: agriculture, protected open space, stewardship and destination hospitality an hour from a concentration of technology, wealth and talent. And a wine economy in real transition.
Puget Sound / San Juan Archipelago. Fisheries and marine food systems alongside farming, islands, protected land, tourism, Seattle nearby. Real personal history: Port Townsend, Seattle, and a deliberate off-season trial of island life on Bainbridge.
Mendocino / North Coast, including Anderson Valley and the coast. The wildcard, and the place Tyler knew least. Remote enough that getting there is friction, and maybe part of the point. High curiosity, low confidence.
Santa Fe / Northern New Mexico. On the list for the singularity of the place: landscape, architecture, climate, history, traditions, ecology, and a relationship between food, land and community unlike anywhere else on the list. No venture thesis required for it to deserve exploration.
Louisville / Kentucky. A different kind of advantage: trust, access, relationships, Rye, and real familiarity. Open question even then: what the relevant geography actually is, Louisville, central Kentucky, bourbon country, or something broader.
Michael’s five archived
Pacific Northwest. The one entry he had explained in writing at the time. His stated reasons: regulatory environment, fertile soil, food culture, and the fishing industry broadening what agriculture means.
Upstate New York / Hudson Valley.
Bay Area.
Great Lakes.
Kentucky.
Beyond the Pacific Northwest these were named but not yet reasoned in writing. The archived hub’s own note: “We are asking, not assuming.”
The place-selection logic, as it stood archived reasoning
The archived hub recorded the pattern it saw in Tyler’s list rather than hiding the bias in it: he did not start from a national scan; he named places he already knew. The stated pattern: places with a strong identity rooted in land and ecology · real food and agricultural culture · meaningful natural assets and a real culture of stewardship · a reason people already want to be there · some relationship to a larger market without being inside it. North Bay and San Francisco; Puget Sound and Seattle; potentially Hudson Valley and New York City. And in several of them, something important also appeared to be changing.
Its own definition, worth keeping: “The landscape matters” here means the physical geography materially shapes what can be grown, how people live, what gets protected and why people travel there, not that the scenery is nice.
Status of the geography hunt
THE BROAD GEOGRAPHY HUNT IS PAUSED AND ARCHIVED, NOT COMPLETED AND NOT DECIDED. Michigan, Washington, New York, New Mexico and the rest are shelved, not abandoned. The active hierarchy is: North Bay = broader regional context · Sonoma County = the primary system currently being mapped · Sonoma Valley = the microscope · Louisville / Kentuckiana = comparison system only, never a build-location evaluation. NO VENTURE GEOGRAPHY HAS BEEN SELECTED. Sonoma is the working lens, not a chosen answer.
C · ARCHIVED Models, references and early research
The earlier hub carried a substantial research universe that this hub does not reproduce. It is preserved whole in the archive rather than summarised away. What is in there:
Landscape section and the interactive map with layers, region bar and region profiles
“Key places in this region” and the county ranking work
County comparison tooling and landscape research across the candidate geographies
Models: hospitality examples, farms and farm stays, food businesses, co-ops, infrastructure examples and institutions
Library: what the project was reading, watching and listening to, with filters and tabs
Evidence section and the Wave 1 field-research findings, September 2026
Research method: “How this research works”
The deeper open questions, Phase 1 progress, and Method, sources & limits
Early thesis language and the early framing around food systems, place, community and hospitality
Deliberately not condensed here. The point of the archive is to preserve the research path, not to describe it. Open the archived hub to work with any of it.
D · THE BRIDGE What changed
This work was not discarded. It led to the current inquiry. As the regional food system was mapped more deeply, the question shifted from “what place-based hospitality business could strengthen a regional food system?” toward “what functions between producer and buyer are actually missing, duplicated, poorly coordinated, or already working well enough, if any?”
What moved to the front: the middle became the lead inquiry. That means aggregation · distribution · market access · buyer procurement · storage · processing · coordination · provenance · demand signalling · producer capacity.
This is NOT a conclusion that a distributor should be built. It is a statement about where the questions currently are. Basic produce aggregation and distribution demonstrably already exists in this region and is retired as a gap. The live question is where existing middle functions stop serving producer or buyer needs adequately, if anywhere, and the honest answer may be “nowhere.”
Standing principles, unchanged through the pivot
MAP BEFORE INVENTING · LISTEN BEFORE SOLVING · FIND THE SEAM
What carried across the pivot
Committed demand. Explored earlier as an owned kitchen buying from producers; now central to the current inquiry as demand aggregated backward into production signals.
Sense of place as an economic asset. Earlier the basis of a hospitality thesis; now a hypothesis about whether attachment to place can support a broader agricultural landscape.
Land. Earlier a property question for a venture; now a candidate constraint on local production itself.
Hospitality. Earlier the venture; now a buyer and demand node, and three October stops.
Strengthening communities through food systems. Unchanged as the outcome the work is for.
Open the full archived hub
The complete pre-pivot hub is preserved unrewritten, with only an archive banner added at the top. Nothing was removed, condensed or reinterpreted.
The original untouched copy also remains as index_ARCHIVED-geography-scan_2026-09-02.html. Neither archive should be edited or overwritten.
Overlays & method notes
kept deliberately at the bottom - the map is the project; these are lenses on it
Rising Tide HYPOTHESIS / POSSIBLE RESPONSE - NOT PROJECT DIRECTION
An overlay, off by default. Switching it on asks one question: which VERIFIED problems would this actually solve? A valid outcome is: none that aren't already solved.
FRONT everyday regional-first market
Maps to F6's open question: is there demand for everyday regional-first retail beyond market hours? Status: unproven in either direction H. Next evidence: grocery local-share audit + demand behavior - not more concept.
MIDDLE shared production + value-add
Maps to F2, the strongest current friction signal (no Valley production kitchen on the county list reviewed to date). Status: the most evidence-adjacent third H - but friction signal ≠ opportunity until utilization and producer demand are known.
BACK aggregation / storage / logistics / B2B
Maps to F3/F5 - where FEED already operates. Status: weakest case for newness duplication risk HIGH. The test is FEED's headroom and limits, not our ambition.
Design principle if it ever earns life: share the expensive functions between producer and buyer, preserve independent operators. Test: does it solve a real need not already met? It may not.
Outside method to watch TransCap / Second Growth
One question matters for us: how do people move from describing a system to deciding where intervention is actually justified? Their transition-map method (explicit leverage criteria, including whether capital can even act on a point) is worth learning from. Their answer is a Midwest answer; nothing here implies it applies to Sonoma. Michael carries the 9/24 session; the hub carries only the method question.
Archive paused, not deleted
The national county scanner (weighted lenses, hotspot map, rankings) that previously fronted this hub is archived: geography hunt paused 9/11; Sonoma is the working lens, not a chosen answer. Local files: almanac-hub_PRE-PIVOT_ARCHIVE_2026-09-17.html (banner added, otherwise unrewritten) and the original index_ARCHIVED-geography-scan_2026-09-02.html, both next to this one on your machine. See Project evolution for the bridge. Michigan / Washington / New York are shelved, not abandoned.