Meet the platform
A producer matching platform built around a simple idea: when the US government stopped its farmer aid programs, years of vetted African producer data was left sitting on government and archive websites, organised by nobody. The platform matches those producers with buyers. Its working prototype ran on demo records. Our job was the data pull: find the real producers, verify them, organise them, load them.
The challenge
There is no single producer registry for African coffee. The data lives in fragments: an organic certification database here, a factory listing there, archive evaluations somewhere else, and several official sources that are simply dead. Some countries publish nothing at all. A matching platform built on scraped guesses would match buyers with ghosts, so every record had to come from a real, named, public source, and every gap had to be admitted rather than filled.
The solution
Three rounds, one sheet, one tab per website
All data went into a single sheet, one tab per source website, so any record can be traced back to where it came from. Round one, on November 15, pulled 962 records: certification lots, winning auction lots, washing station lists, exporter directories and certified company rows. Round two, on November 16, added 1,369 records in 13 new tabs, the biggest being 761 records from the USDA organic integrity database, an official full database export. Round three, on November 17, added 1,787 records in 6 tabs, including 983 Kenya factory listings, the single biggest tab in the build. The arithmetic reconciles exactly: 962 plus 1,369 plus 1,787 is 4,118.
A targets tab that says what everything costs
The targets list grew from the original 43 sources to 80 mapped sources, and gained sector, cost and direct download columns for every row: government, private or mixed; free, free account, paid or request only. Anyone continuing this work can see at a glance what is pullable today and what is gated, instead of rediscovering it.
The farmer reports, verified link by link
The same sheet holds all 27 verified coffee farmer reports from the USAID DEC collection. The original host for those reports is dead, so working copies were found in archive collections, and every single link was curl verified returning 200 before it was added. One report was found and deliberately not added: both its archive page and its PDF return 404, and a dead link in a verified list is worse than a gap.
The gaps, stated plainly
Ethiopia has no public government producer registry, licensed exporter list or washing station database: the official sites are dead or down, and only private directories and union level records could be pulled. Kenya licensing moved to a revived Coffee Board under a 2026 act with no register published yet, and its biggest listing source is a non official compilation that must be verified before use. South Sudan has almost no public coverage at all: 9 named farmer organisations from one USAID report is everything that exists publicly. Several sources stay blocked behind logins, paid API tokens and request only lists, and they are logged as blocked, not silently skipped.
The results
- 4,118 producer records in one organised sheet, one tab per website
- 3 rounds across November 15, 16 and 17, counts independently read back after each round
- 80 sources mapped with sector and cost for every one
- 7 countries covered, Kenya added as the seventh during the build
- 27 verified farmer reports, every link checked live
- $0 spent on logins, subscriptions or paid data
The return is the asset a matching platform actually runs on: real, sourced, buyer ready producer data that cost nothing to access because it was public all along, just scattered. The expensive part was never the data. It was the organising, and that part is done.
FAQ
Where did the producer records come from?
Public sources only: government databases, certification bodies, auction and factory listings, exporter directories and archive collections. Every tab in the sheet is one website, so every record traces back to its source.
Why one tab per website instead of one big list?
Provenance. A merged dump loses where each record came from, and with it any way to verify, refresh or defend the data. One tab per source keeps the audit trail intact.
What could not be pulled?
Ethiopia publishes no public producer registry at all, several sources sit behind logins or paid tokens, and one farmer report exists only as a dead link. Those are logged as gaps in the sheet rather than filled with guesses.
Are the Kenya listings official?
No, and the sheet says so. The largest Kenya source is a non official compilation, flagged for verification before buyer use. Flagging a source honestly is part of the deliverable.
