grading method
Data Quality (DQ) ratings.
Tested against the source on six counts.
DRAFT · no ratings are issued yet · feedback welcome at hello@tollfair.com
six dimensions
What we test.
- 01weight 20%
Provenance
Is the publisher the original holder, and can every record be traced to it?
- 02weight 15%
Freshness
How long after a change at the source does it appear in the API?
- 03weight 25%
Accuracy & validity
Do sampled records match the source field by field, in the documented formats and schema?
- 04weight 15%
Completeness
What share of expected records and fields are present?
- 05weight 10%
Licence clarity
Can an agent tell, from the response alone, what it's allowed to do with the data?
- 06weight 15%
Reliability
Uptime, latency, error rates and unannounced schema changes, measured from outside over time.
the scale
A score out of 100, banded A to E.
Each dimension is scored 0–100. The DQ score is their weighted average, shown as a band you can read at a glance, like an energy label: DQ A · 88.
- DQ A85–100Excellent. Dependable for unattended agent use.
- DQ B70–84Good. Minor, documented gaps.
- DQ C55–69Fair. Usable with checks on the weaker dimensions.
- DQ D40–54Weak. Significant gaps found in testing.
- DQ E0–39Poor. Not fit for unattended agent use.
- A Licence clarity or Provenance score below 50 caps the rating at C, however strong the rest is.
- A rating expires 90 days after its last test and shows as Ungraded until it's retested.
ground rules
Tests decide. Nothing else.
Only tests move a rating
Every score comes from tests anyone could rerun against the source. Nothing else changes it.
Stars measure demand
Stars decide which datasets we test first. They never feed into a score.
Reports trigger tests
A credible evidence report queues a retest of that dimension. The report itself changes nothing.
Ratings show their age
Each rating carries the date of its last test, and expires if it isn't retested.