The Bulletin
Winning French public contracts that never reach TED
The national French procurement bulletin publishes the contracts below the European thresholds — the ones TED never carries. It's free, it's public, and almost nobody works it systematically. The play is to read both sides of it: the buyer with an open tender you can bid on, and the supplier who just won one you can partner with.
Who the campaign is aimed at
French public buyers with a live tender, and the suppliers who just won one. Two sides of the same bulletin, worked as one pipeline.
The segments it splits into
Open tender — bid now
The notice is a live call for bids with the response window still open. The move is to reach the buyer while there is still time to respond, or to be on their radar for the next one.
Attribution — who just won
The notice is an award: a supplier just won a public contract. A proven public-sector seller — a partner to subcontract to, a competitor to watch, or an account that now has budget to spend with you.
Read the count before the pull
The free preview returns a filter summary, a count and the first ten notices without writing anything. It is the cheapest way to find out whether a query is aimed right — too broad, too narrow, or exactly on the contracts you can actually win.
Treat it as the tuning step. A query refined on a zero-cost preview is a query that imports clean, instead of a pull you spend the next hour deleting.
One company per public buyer
The pull imports the buyer behind each matching notice, with the tender attached as a durable signal: the buyer's legal type and sector, the framework and procedure, the amount, the duration and the notice URL. Filter it on keywords, on family — national FNS or European JOUE — on the type of contract, and on the départements you can actually serve.
Every notice lands as a public_tender signal, so the reason to reach out travels with the account. You are never holding a buyer without the contract that made them worth contacting.
Read who won, not just what's open
Flip the nature of the query from call-for-bids to attribution and the same bulletin hands over the other side: the suppliers who just won. That is Profile B — companies that have proven they sell to the public sector, which is a partner shortlist, a competitor watch and a subcontracting angle all at once.
The two passes share every other filter, so one campaign covers both the demand side and the supply side of the exact same market.
The lot is in the detail
The list view is a summary; the decision is in the parsed notice. Pull the full record for the shortlist and read the object, the procedure, the real amount and the response terms before anyone spends time on a bid or an email.
This is where a promising line reveals itself as a framework you can't join, a lot far below your floor, or exactly the contract you should be responding to this week.
Winnable, not just relevant
A tender that mentions your keyword is not a tender you can win. This pass asks the harder question per notice: is this inside your capability, above your minimum, in a region you serve, and open long enough to respond to at all?
It runs on the shortlist, not the raw pull, because it is the most considered read in the pipeline. Everything that clears it is a contract worth a human's attention.
Say no to the unwinnable
An explicit gate: out go the lots below your threshold, the procedures you can't enter, the départements you don't cover, and the frameworks already closed to new entrants.
The discards are read, not dumped. If a whole descriptor keeps importing contracts you can never win, that is a keyword to drop at the pull, not a rejection to repeat every week.
The name the buyer published themselves
The last step reads the contact point off the notice itself — the name, email and phone the buyer published — and writes it as a contact. No enrichment guesswork: this is the person the buyer put on the record as the one to talk to.
It only works on notices this pipeline sourced, because it needs the notice reference to read from. On the smallest lots the buyer published none, and the honest answer there is not_found rather than a fabricated address.
The smallest lots often carry no published contact at all — expect a share of not_founds and treat them as normal, not as a failure. A guessed generic address on a public tender is worse than no address.
Where the list thins out
The numbers below are a worked example, not a single client's run: a quarter of the bulletin on one sector's keywords, put through the seven steps. The ratios are the argument — and the last stage is the honest one, where the smallest lots simply carry no contact.
| Stage | In | Kept | Survival |
|---|---|---|---|
| Notice pull | 2,400 | 2,400 | 100% |
| Relevant to offer | 2,400 | 820 | 34% |
| Read in full | 820 | 540 | 66% |
| Winnable | 540 | 310 | 57% |
| Buyer contact published | 310 | 230 | 74% |
The bulletin is free and public, which is exactly why almost nobody works it end to end — reading two thousand notices by hand is nobody's idea of a week. Run it as a pipeline and it reruns on every fresh batch, on both the open and the awarded side, for the cost of nothing but the qualify pass.
Want to run this play on your market?
Every Walegold recipe is built and run on Walego — the sourcing, the enrichment, the profiling and the founder hunt. Bring your ICP and we'll build yours.
Try Walego