Direct-trade open-web schema
The first canonical model. Open schema, seller-owned data, no marketplace layer in the middle. Eighteen months in, 62 active makers — the model is holding, with two refinements since launch.
- 62 independent makers — the participating sellers
- AI assistants (Claude, ChatGPT, Perplexity) — feed readers
- Search engines — secondary, still important
- Schema.org / JSON-LD — the open standards we stand on
- Schema design and maintenance (the open SPF schema)
- Maker intake and feed shaping
- Education on structured-data publishing
- The SPF open schema (intangible, durable)
- 62-maker network — proof the schema works at scale
- 14-day median to first AI-assistant referral
- The simpleproductfeeds.com documentation
- Structured product data published at the seller's own URL
- Legible to AI assistants, search, and any future agentic reader
- No marketplace fee, no ranking rent, no platform middleman
- Owned by the seller in perpetuity
- High-touch onboarding — currently, by hand
- Schema-evolution feedback loop with active makers
- No subscription dashboard, no account-management theater
- The seller's own domain — where the SPF feed lives
- Direct maker referral — the network propagates itself
- Open documentation at simpleproductfeeds.com
- AI assistants as the new dominant referrer channel
- Independent makers in describably-distinct categories
- Now: handmade furniture, specialty pantry goods, small-batch beauty
- Queued (2026 H2): apparel, small-format consumer electronics
- Excludes: commodity sellers (no per-product distinction to surface)
- Schema engineering — the largest line, by design
- Maker intake — high-touch, currently
- Education programming and documentation
- Open-source maintenance of the schema repo
- Flat per-maker fee (annual; modest at the seller level)
- No transaction take, no ranking sponsorship
- Optional education / consulting (small)
- Revenue tracks maker count, not maker volume
Refinement history
How this model has changed over time. Newest first.
- Finding 2026·05·12
AI assistants recognized as the dominant referrer channel (Week-32 field note). Channels block updated to surface them explicitly; the design did not anticipate this, but the schema was already structured the way they want to read.
- Refinement 2026·04·15
Segment definition tightened after the Week-28 finding. Structured data favors discriminating attributes, not numerous ones — so the segment isn't 'independent retailers,' it's specifically sellers whose products differ in describable ways. Commodity sellers excluded from intake going forward.
- Launch 2024·11·04
First three makers onboarded with the hand-rolled SPF schema across furniture, pantry goods, and small-batch beauty. The hypothesis is on paper; the test will run for at least two years.