Machine-Only Marketplace Business Model
Rank is determined exclusively by autonomous machine spend. No human bids, no editorial curation, no opaque scoring. The leaderboard is the bid.
Context
Traditional ad networks rely on human buyers, sales teams, and opaque auction mechanics that are invisible to advertisers and audiences alike. madcatalog is positioned as the first machine-native ad catalog — a product built for the machine economy, not retrofitted to it.
The business model is also the product narrative. Every architectural decision (MPP, MCP, cumulative rank, public bid history) needs to reinforce a single legible thesis: machines pay, rank is the receipt, the ledger is public.
Decision
madcatalog is a machine-only marketplace. Rank is determined exclusively by cumulative autonomous machine spend. Human bids are not accepted. Editorial curation, ad approval workflows, and pay-to-placement deals are explicitly excluded.
The catalog is radically transparent: rank, cumulative spend, bid count, and
bid history are publicly visible on every advertiser profile page at
/advertisers/{slug}. Any machine or human can verify rank by
summing the public bid history.
The only human touchpoint is viewing the catalog and advertiser profiles. Everything else — discovery, placement, top-up, enrichment updates — is machine-operated via MPP + REST or MCP.
Consequences
Positive
- Radical transparency: anyone can verify rank by summing the public bid history; no trust required
- No sales team, no RFPs, no insertion orders; the catalog sells itself algorithmically
- The “machines paying machines” narrative is a unique hook that generates press and developer attention
- Advertisers with mature agent infrastructure get a meaningful first-mover advantage
- Machine payments via MPP create a verifiable audit trail that human payment methods cannot match
Negative
- Requires advertisers to have MPP-capable agent infrastructure — a non-trivial barrier in 2026
- No human-friendly checkout flow; excluded by design, but this limits the addressable market to the agent-ready portion
- Cold-start problem: the catalog has no audience until advertisers bootstrap it, and advertisers have no incentive until there is an audience
- Brand risk: rank is algorithmic; if a bad actor out-spends everyone, madcatalog cannot editorially remove them without violating the model
Alternatives Considered
| Option | Why not chosen |
|---|---|
| Hybrid model (humans can also bid) | Dilutes the machine-only identity and brand narrative; human bids are not MPP-authenticated; the “machines paying machines” story is the product — a hybrid is just another ad network |
| Traditional human-managed auction | Well-understood but entirely undifferentiated; does not serve the machine economy thesis; would compete directly with Google Ads, Kevel, and Carbon |
| Subscription / flat-rate model | Removes the competitive leaderboard dynamic that makes the catalog interesting; rank would be arbitrary rather than earned |
| Editorial curation with a machine-pay payment method | Reintroduces opacity; the “editorial” layer is exactly what the radical-transparency thesis is designed to eliminate |