June 2, 2026 · 6 min read
Why we ship a free tier with 1,000 API calls/month
Most travel APIs are gated behind sales calls, trial requests, and a procurement quote you'll get back in nine business days. Ours isn't. Sign up, copy a key, make a call. The free tier is permanent — 1,000 calls a month, no card. That's a deliberate choice.
Two reasons, both about conversion
We picked the free-tier shape for two related reasons. The first is that a chatbot prototype needs to be testable in ten minutes, not ten days. The second is that for products in our price band, a self-serve trial converts to paid at a much higher rate than an enterprise demo converts to a contract.
The ten-minute test
Imagine you're four weeks into building a travel chatbot. The LLM is hallucinating Indian destinations. You search for a fix and land on our home page. You have an hour before standup. What you want is to paste a curl command into your terminal, see real data come back, and judge — in less than ten minutes — whether this would solve your problem.
A "request a trial" form does not let you do that. It puts a salesperson between you and the answer. The salesperson is helpful in some markets, but for a developer evaluating a data API, they are friction. By the time the trial is approved, you've shipped a worse fix and moved on. The deal is lost before it started.
Why 1,000 is the right number
A thousand calls a month is enough to build a real prototype, run it through manual QA, and demo it to a teammate. It is not enough to run a production chatbot. That is exactly the design intent: free is for evaluation, paid is for production. The line is clear, the customer doesn't feel tricked, and the upgrade conversation is obvious — your prototype works, you have users, you need more calls.
We considered 100, 500, and 5,000. A hundred is too few; a serious test plan exhausts it before the developer is convinced. Five thousand is generous to the point of subsidizing real production use on the free tier. A thousand sits at the convex point of the curve.
Self-serve conversion math
For products priced at $49 to $249 a month, the well-known industry pattern is that self-serve free-trial-to-paid conversion sits in the 3-7% range, while enterprise demo-to-contract conversion is much lower in raw count even if the per-deal value is higher. For our customer profile — small SaaS teams building chatbots — the volume is on the self-serve side. The economics agree.
Even more importantly, self-serve customers cost almost nothing to acquire after the page is up. There's no AE, no proposal cycle, no custom contract redlines. The cost of serving a free user is the marginal API call cost, which for our shape is fractions of a cent.
The signup flow
The flow we ship is intentionally short. Land on
/app/signup.html. Enter email and a password. Confirm
via emailed link. Land in a dashboard with one section: your API
key, a copy button, and a curl example. A first call works
instantly. No card, no demo call, no sales touch. Total time from
intent to first 200: well under five minutes for someone who reads
the docs.
The dashboard later grows usage charts, plan management, and a Paddle-checkout upgrade button. None of that gets in the way on day one.
What we'd change if we had to
The honest answer: nothing yet. The free tier has been the single biggest source of paid conversions in the first months. The skeptical version of me wonders whether we'll need a card-on-file free tier eventually for abuse prevention, but the current shape is working. Don't fix what isn't bleeding.
If you're shipping a B2B data API in this price band, ship a free tier. Make it permanent. Make signup take five minutes. The customers who want to talk to sales will still find the contact page; the customers who want to ship next Tuesday will thank you.