Research notes / 001 · 2026-09-02
Fewer notices.
Not fewer questions.
We tested the complete company-scan workflow against real TED data. Here is what the test supports—and what it does not.
500
Notices considered
13
Passed first filters
3
Reviewed in depth
What we measured
A bounded pool of 500 ingested notices was considered for one company profile. The configured first-pass rules excluded 487 notices (97.4%). The remaining 13 passed those filters. Passing them does not prove legal eligibility.
The test reviewed only the highest-ranked 3 candidates in depth. That was a review cap, not a finding that the other ten were unsuitable. We did not measure whether the system missed better opportunities outside this pool.
The result was three HOLDs. Not three approved bids.
Fit scores were 71, 70, 70 out of 100. All three needed checks against the full procurement documents. The retrieved descriptions did not establish mandatory eligibility requirements, and company delivery or language evidence was incomplete. A strong-looking match was not upgraded to GO without proof.
| Priority / source | Fit score | Decision |
|---|---|---|
| 1. TED 605921-2026 | 71/100 | HOLD |
| 2. TED 605563-2026 | 70/100 | HOLD |
| 3. TED 605472-2026 | 70/100 | HOLD |
What we have not measured
This test does not establish customer hours saved, money saved, bid win rate, legal eligibility, or completeness across the EU tender market. There was no controlled comparison with a human bid team. We do not convert the excluded-notice count into claimed time savings.
The homepage calculator is a separate, editable scenario. Its default time and hourly-cost inputs are illustrative assumptions, not observations from this test. It can return a negative value. A reduction in review time does not necessarily reduce payroll or spending.
Where facts come from
TED publishes procurement notices and provides a search API for retrieval and reuse. Read the Publications Office’s TED API documentation. TenderMeter retains notice IDs and source links. Buyers’ official documents, later amendments, and submission instructions always prevail.
The next evidence milestone
A prospective pilot should record the same team’s review time before and with TenderMeter, keep comparable tasks and opportunity counts, and record false positives and missed opportunities. Until those measurements exist, we will keep customer-saving claims off this page.
Discuss a measured pilot or explore the qualification process.