Why it matters
- Tune your models. A dip after a model change is immediate feedback that something regressed.
- Calibrate confidence. Compare acceptance across confidence bands — if your “90+” proposals are rejected as often as your ”50s”, your confidence scoring needs work.
- Prove value. A high, stable acceptance rate is the case for expanding scope with the merchant.
Endpoint
catalog:read scope. The organization is resolved from your API key, so the metrics are
always scoped to your own submissions for that merchant.
Optional. Restrict the window to proposals reviewed after this moment (epoch millis or ISO-8601),
mirroring
modifiedSince on the products endpoint. Proposals not yet reviewed are matched by their
submission time instead, so the counts always add up.Optional. Restrict the window to proposals reviewed before this moment.
Response
All proposals you’ve submitted in the window, regardless of review state.
Proposals a reviewer has acted on (accepted or rejected).
Proposals where at least one field was applied to the catalog. A proposal you review field-by-field
and partially approve counts as accepted.
Proposals a reviewer rejected in full.
Proposals still awaiting review.
accepted / reviewed, as a 0–1 fraction rounded to four decimals. pending items are excluded from
the denominator; the rate is 0 until at least one proposal has been reviewed.The same
reviewed / accepted breakdown split by the confidence you sent — the fastest way to
check whether your confidence scoring is well calibrated. Proposals submitted without a confidence
are counted in the totals above but not in any band.