how to price an mcp server (pay-per-event, with real numbers)
MCP server pricing on apify is pay-per-event: you define named charge events, typically a start event at a fraction of a cent plus a per-result event at $0.05 to $0.10, apify meters and bills the buyer, keeps 20%, and pays you the rest. once set, a price locks for one month. those are the mechanics from my own 19-server catalog, not a hypothetical.
this is the pricing spoke of how to sell mcp servers on apify. everything below is verified against my live account, and every trade-off cost me something to learn.
the three pricing models, compared
| model | how it bills | best for | watch out |
|---|---|---|---|
| free | nothing (buyer still pays apify platform usage) | funnel actors, discovery, list-building | going paid later still hits the one-month lock |
| monthly rental | flat fee per month | tools with continuous, predictable use | a hard sell for tools an agent calls occasionally |
| pay-per-event | per named event: start, result, custom events | MCP tool calls, exactly the agent usage pattern | every pricing entry needs the 20% margin field; updates are append-only |
for MCP servers, pay-per-event is the natural fit because agent usage is bursty: forty calls during one deal review, then silence for a week. nobody subscribes for that; anybody will pay sixty cents for it.
my actual price list
- $0.10 per result: compliance scaffolding (EOR, EMEA, APAC, LATAM, EU AI act), MEDDIC coaching, SEC 10-K excerpts, tech-stack interpretation, SaaS pricing benchmarks, buyer signals, DM scripts, sales engineering
- $0.05 per result: lighter tools: SMB intelligence, hook formulas, rate calculations, pipeline hygiene
- $0.001 per start on everything paid
- free: mutual action plans, multi-threading gaps, MSA redline stances
the logic: price tracks the value density of one call. one compliance answer that keeps a deal moving is easily worth a dime; a hook formula is a nickel. and the number stays low enough that no procurement conversation ever happens, which for agent tools is the entire game. the catalog itself is mapped in the sales-team post.
why some of my servers are deliberately free
three of my 19 are free on purpose, and it's not generosity, it's routing. a free server is a discovery surface: it ranks in the store, costs nothing to try, and its README walks the reader straight to the paid catalog. my mutual-action-plan server is free while the MEDDIC coach that contains a superset of it is paid. the free one is the doorway.
this mirrors how my whole site works: the rate calculator is free, the AI visibility scan is free, and the paid offers sit behind demonstrated value, never in front of it.
the one-month lock changes how you launch
apify rejects any pricing change within one month of the last one, including switching to free, and paid actors can't quietly unpublish either (you must schedule them free first and wait). so the sequence that works:
- verify the server end to end BEFORE the first pricing entry, because you can't cheaply walk it back
- set the price you can defend for a month, not the experiment you'd like to run for a week
- publish only after both, and mind the store's 5-publications-per-day cap when launching a batch
pricing an MCP server is closer to pricing a product than tuning an ad. that's not a complaint; the lock forces the discipline of deciding what a call is worth before asking anyone to pay it.
what the 20% buys you
apify's margin on my actors is 20% of billed events. for that, i never touch metering, invoices, payment failures, fraud, tax handling on payouts, or hosting. as a one-person operation, replacing all of that with stripe and my own infrastructure would cost far more than the margin does. the full build-vs-buy reasoning is in apify vs self-hosting.
frequently asked questions
how does pay-per-event pricing work?
you define named charge events in the actor's pricing: a start event and a primary per-result event are the common pair. apify meters them, bills the buyer's account, and pays you out minus 20%.
what should my first price be?
the one you can defend for a month. for framework and data tools, the working range in my catalog is $0.05 to $0.10 per result. low enough to skip procurement, high enough to be a real business at agent-scale call volumes.
does a price guarantee revenue?
no. a priced listing with no distribution earns nothing; my own catalog's organic usage rounds to zero so far, and i say so in the pillar post. price sets the value of a call; distribution decides whether calls happen.
the pricing discipline here is the same one in my AI builder toolkit at closermethod.com/skills: decide what the output is worth, prove the value free first, charge for the compounding version.
go deeper
- how to sell mcp servers on apify (19 live, the honest version)
- the full catalog: 19 mcp servers for sales teams, mapped by deal stage
- how to deploy an mcp server on apify that survives the daily test
- apify vs self-hosting your mcp server: the real trade
sources: apify actor monetization documentation; all prices, the 20% margin (apifyMarginPercentage on every pricing entry), the append-only pricing API behavior, the one-month change lock, and the unpublish restriction verified against the live apify account and API responses, may through july 2026, most recently july 27, 2026. written by elisabeth hitz, certified in anthropic's ai fluency program (framework & foundations, and ai capabilities & limitations), plus claude 101 and claude cowork.