How we count public bids
The rules behind every number we publish: what counts as a bid, what we leave out, and where our data falls short.
If you cite a VelocityBids number, this page explains what it measures and where it falls short.
What we collect
We monitor 20,000+ public procurement sources: state portals, city and county sites, school districts, universities, and special districts such as transit, water and housing. Most are collected once a week on an assigned weekday. Many sources are quiet at any given moment:
| Sources | Count | As of |
|---|---|---|
| Active | 26,121 | November 1, 2026 |
| Active, leaving out resellers | 24,883 | November 1, 2026 |
| Returned listings in the weekly run | 15,513 | October 29, 2026 |
| Gave us a new bid in the last 30 days | 9,626 | November 1, 2026 |
| Holding an open bid right now | 4,695 | November 1, 2026 |
We added 2,331 sources between October 29 and November 1, so the first row is newer than the third.
What counts as a bid
- One row per solicitation. Duplicates we detect across sites are merged, so a bid on two portals should count once.
- State and local unless we say otherwise. Federal figures from SAM.gov are kept separate.
- US only. Non-US sources are excluded.
- No resellers. Aggregator listings are excluded, so we count what buyers published.
- Two portals are left out of our weekly bid counts and time-to-respond figures because their terms limit how we can use their data. Our document shares, coverage tests and market estimate were measured with their rows included.
Dates
Weekly counts use the due date, not the day we first saw a bid. Counting by first-seen date inflates any week when we add sources. Time-to-respond figures use only bids with a posting date from the source more than a day before our first sighting, and they drop windows of 120 days or more.
How much of the market we see
We estimate we hold 45% to 70% of state and local solicitations, with a central estimate of about 58% and medium-low confidence, as of October 29, 2026. No single test gives that number. We ran four checks of whether bids found outside our system were in our data:
| Test | Bids checked | Share we held | Main weakness |
|---|---|---|---|
| Web sweep, exact match | 236 | 44.5% | One company's product categories; exact matching is a floor |
| Web sweep, recent bids | 199 | 46% exact, 53% fuzzy | Same categories |
| Recheck of open bids, October 29 | 93 | 70% | Small sample |
| Comparison with a large aggregator | 13,803 | 31% to 38% | Biased low: those are the buyers we reach least |
The first three are small and cover one set of categories, so they may not hold across every trade. We treat the fourth as a floor for buyers who post mainly on that aggregator. Coverage changes as we add sources, and we will update this table when we re-test.
Documents
We hold a document for 51% to 58% of state and local bids, depending on whether you measure captured flow (51%, October 29) or open bids (58%, November 1). For federal bids on SAM.gov the figure was 29% on October 29.
Known weaknesses
- Very short bids are under-counted, because most sources are collected weekly.
- We have not measured precision. Some rows may be headings or archive items instead of true solicitations.
- Trends are not market trends yet, because our source base is still growing.
- State figures reflect the buyers we cover, which is not every buyer in a state.
- We do not publish trade figures yet, because our category labels are still being built.
What we publish
Our statistics are aggregates. We do not publish customer or prospect information, and these articles do not reproduce bid listings.
Using our numbers
You may cite these figures with a link to the source article. Send questions or corrections to hello@velocitybids.ai.