Where state and local RFPs actually live

Most people start with SAM.gov, find it overwhelming, and conclude that public procurement is a federal game. Then they win a school district contract by accident and realise the money closer to home was never on SAM.gov at all.

Below the federal level there is no central index. Every state, county, city and university system publishes on its own system, and those systems fall into a small number of families. Knowing the families is most of the work.

The platform families

Bespoke state portals. A handful of states run software written for them. Wisconsin’s VendorNet is the one we know best. These are usually the richest listings, with real deadlines and attachment lists, and they are also the most likely to break when the state does a redesign.

Periscope / BuySpeed. A shared procurement platform that a number of states sit on. The listing pages look almost identical from state to state, which sounds convenient until you try to read one programmatically: the search results come back from a form post with a session token rather than from a URL you can link to. You cannot bookmark a search. You often cannot link to an individual bid.

Jaggaer. Common across university systems. Each institution is a separate tenant with its own subdomain, so a single university system can mean a dozen addresses to watch.

ionWave. Widely used by cities and counties. Small listings, plain HTML, easy to read, and enormous in aggregate because there are so many of them.

Grants.gov and SAM.gov. Federal, well documented, and the only two with APIs a normal person can call. Everyone has these. They are table stakes, not an advantage.

Why searching across them goes wrong

Three specific things, and none of them are obvious until you have tried.

There is no shared classification. Federal grants use CFDA numbers. Federal contracts use NAICS. Most state and local portals publish no industry code at all, or use a home-grown commodity code that means something only inside that state. If you build a search around NAICS because that is what you know from the federal side, you will get an empty result set from most of the country and read it as “nothing this week.”

Deadlines are inconsistently published. A surprising share of listings carry no closing date in any machine-readable field. It is in the PDF, or in a sentence in the description, or nowhere. If you sort a combined list by deadline, every listing without one vanishes from the view. We did this to ourselves early on and hid most of our own corpus for a while.

Nothing notifies you reliably. Some portals offer email alerts. The alerts are usually scoped to their own commodity codes, so you either get everything or nothing, and they stop silently when a registration lapses.

What to do about it if you are doing this by hand

Pick the jurisdictions where you have actually won work, or where you have a credible reason to be competitive, and ignore everything else for now. Breadth without focus produces a list nobody reads.

For each one, find the listing page rather than the homepage, and check whether the URL changes when you search. If it does, you can bookmark a filtered view and check it weekly. If it does not, you are stuck re-entering the search every time, and that portal is the one you will quietly stop checking in about six weeks.

Put the check on a calendar with a named owner. “Someone looks at the portals” is how a bid closes on a Monday nobody looked.

Keep a note of what you passed on and why. The same solicitations repost on a cycle. Twelve months from now you want to know whether the bonding requirement that ruled you out last time still applies, and you will not remember.

What we built instead

RFP Docket exists because we got tired of doing the above and doing it badly. It reads these portals on a schedule, including the ones that need a real browser to drive, and puts the matches in one list. The coverage page says exactly which sources run today and which are still in build, including the two that need rebuilding because the states left BuySpeed.

The part that surprised us: the aggregation was never the hardest problem. Keeping the paperwork straight after you find something is worse, and almost nobody builds for it.