70K government bids from 605 agencies, $99 MRR — showing my work
Just set up ProcureTap here. Quick backstory + numbers for anyone doing similar.
Origin: I was at a hockey game with a contractor buddy. He spent half the night bitching about the software he uses to find government bids — a mess of janky websites with bids often hidden behind crazy expensive subscriptions, no real search. In the third period he said: "you know computers — can't you build a better version of all this mess?"
Six months later, here we are.
What it does: aggregates every US federal, state, and local government bid (plus international — EU, UK, Canada, World Bank, UNDP, AusTender) into one searchable place. 70K live opportunities, 605 agencies, refreshed every 6h.
Stack: User facing infra is Cloudflare Workers/D1/R2/Pages(SvelteKit). Back end infra is a small fleet of Mac Mini's hosting mysql and the agents that fetch bids. ~$40/mo infra cost, no VC.
Where I am:
- 197K total bids indexed, 70K currently open
- 605 organizations covered (all 50 states + federal + international)
- $99 MRR (1 paying customer, 1 in trial)
- Bootstrapped, solo
What's worked so far:
- SEO first — state hubs + industry hubs with real content. GSC impressions ~8x since February.
- Free tier with unlimited browsing (no card required). Pre-pivot paywall was killing signups — 0/7 trial signups converted before I ripped it out.
- Markdown-for-Agents — every page serves fact-dense markdown to any request with Accept: text/markdown. AI agents can consume the data cleanly, humans get the normal HTML.
Launching on Product Hunt soon. Happy to answer anything about scraper economics, government data compliance, SEO strategy, or the CF Workers stack.
Hi George! Really interesting project—congrats!
I'm building in the public procurement space as well (in LATAM), I am curious about your go-to-market strategy. How are you approaching SEO, and where do most of your qualified leads come from?
Would love to hear a bit about what's been working for you. Thanks!
Back in the glory days of SEO it was about matching keywords. Now that AEO is a thing, I can worry less about the specific keywords and more about the actual semantic intention of the question / answers. That mind shift has help guide all the content decisions. Frankly, it makes way more sense to worry about the message than it does the words.
Removing the paywall fixed a conversion problem, but it also changed what your early traction is measuring. Now the challenge is understanding whether users are finding valuable opportunities or simply finding more opportunities. Those lead to very different product decisions.
My first reaction is always valuable data is worth paying for. But if I relax a bit, I can see that just having the data is only the start of the value ladder. Making it easy to find, easy to understand, and easy to monitor, that adds the real value worth paying for.
Exactly.
One thing I'd keep watching is whether users naturally move from exploring the data to relying on the product for a recurring decision.
That transition usually tells you whether the product is becoming a workflow or remaining a resource people check occasionally.