EU public contract expiry radar
The re-tender is 12 months out. The expiry date is already public.
Tender alerts fire when a notice opens, by which point the specification is written and the outcome is largely decided. An expiry date sits a year upstream — in the window where a challenger can still shape requirements.
A row tells you what it does not know
Provenance is a column, not a convention. Every lead states which of its facts the buyer published and which were computed, so what you are acting on is legible before you act on it.
84% of rows carry an end date the buyer published. The rest are computed from a published start date plus a published duration, and are never presented as published facts.
A supplier is known for 92% of rows. 52% name the holder of that specific lot; the rest name suppliers on a multi-lot award without publishing which holds which.
12,471 contracts sit on frameworks that rerun competition during their term. Those are winnable without waiting for the expiry date at all. See them →
Where the inventory sits
| Sector | Contracts | Value published | Share of corpus | |
|---|---|---|---|---|
| 33 Medical | 36,030 | 46% | → | |
| 90 Environmental | 31,722 | 28% | → | |
| 45 Construction | 31,346 | 20% | → | |
| 71 Architecture & engineering | 29,854 | 27% | → | |
| 50 Repair & maintenance | 13,699 | 42% | → | |
| 60 Transport | 12,843 | 25% | → | |
| 79 Business services | 12,206 | 43% | → | |
| 72 IT services | 8,880 | 42% | → | |
| 85 Health & social | 3,462 | 39% | → | |
| 48 Software | 3,188 | 35% | → |
By country
Or skip the website entirely
The same corpus is a tool. Point an MCP or A2A client at the endpoint and query it directly — six capabilities, identical on both protocols, with discovery free and unauthenticated.
Recompete is paid by vendors for delivered leads, never for outcomes. No fee is contingent on winning a contract. Ranking is computed from published procurement data alone and is not influenced by who pays.