A price monitoring proxy scorecard should judge whether public price records can be replayed, not just whether requests succeeded. The audience is pricing, data quality, and crawler reliability teams; it fits public product pages with visible fields, not private sources or datasets without stored evidence.
Replay status belongs beside every price field
Price, currency, stock, and promotion fields are useful only when analysts can see where they came from. Store the source snapshot, proxy exit, target market, parser status, and replay result with the record.
This lets teams separate a real public page change from a parser issue, region mismatch, or temporary lane pressure.
The scorecard should trigger queue changes
A score without an action is easy to ignore. Use thresholds that move a lane into normal collection, slower pacing, replay review, or temporary hold.
| Scorecard field | Healthy signal | Queue response |
|---|---|---|
| Replay pass rate | Same-market replay confirms fields | Keep current pacing |
| Field completeness | Required price fields are present | Send missing records to review |
| Region match | Target market and proxy exit align | Split or slow the lane |
Cost should include review and retry work
Proxy cost looks lower when it is measured by raw request count. It becomes clearer when measured by usable public price records after replay failures, missing fields, and region mismatch are removed.
That calculation gives pricing teams a better way to compare datacenter proxy lanes, rotating residential proxy lanes, and mixed lane plans.
Keep score windows separate after changes
After changing pacing, retries, or proxy lane design, keep the old and new score windows separate. A merged window can hide whether the change helped.
The lane is healthier only when replay pass rate, field completeness, and cost per usable record improve together.
FAQ
What should a price monitoring proxy scorecard measure first?
It should measure replay pass rate, required field completeness, region match, parser status, and cost per usable public price record.
Why is request success not enough for price monitoring?
A request can succeed while price, currency, stock, or promotion fields are missing, so the record may still be unusable for analysis.