Reference
Measure company matching and technology detection separately.
Test company matching, technology detection, and evidence interpretation separately. A returned row can fail at any of those layers.
Example evaluation record
One example record
- Original website
- example.com
- Company match
- Confirmed
- Technology result
- Supported by two evidence types
- Review outcome
- Accepted
Start with known answers
Use companies your team can verify
A useful test set includes more than easy positive examples.
- 1Known technology users
- 2
Known non-users where the required public checks can complete
- 3
Ambiguous brands, subsidiaries, redirects, and multi-domain companies
- 4
Blocked, incomplete, and intentionally difficult websites
Separate the errors
Do not hide different failures inside one accuracy number
Each error class points to a different product issue and a different consequence for your team.
- 1Wrong company or website match
- 2Missed public evidence
- 3Incorrect detector interpretation
- 4
Incorrect handling of incomplete or contradictory evidence
Evaluation outcome
Inspect false positives, false negatives, and unresolved cases
A practical evaluation should show which findings hold up, which fail, and where human review remains necessary.
- 1
Record why each disputed result was accepted, rejected, or left unresolved.
- 2
Compare errors by field and evidence type rather than only by record.
- 3
Set use-specific thresholds before automating a decision.