What a scraper success rate actually tells you.
A public 30-day success rate is a starting signal, not a data-quality guarantee. Learn how to read it alongside freshness, sample size and field completeness.
The short answer
A public success rate measures the share of runs that completed over a window, as reported by the platform. It does not measure whether the rows were correct, complete or current. Use it with the schema, the last update date and a small completeness test.
Catalogue and method checked 2026-11-12. See how tools are scored.A 99 percent success rate can sit on a tool that returns half the fields you need. A 70 percent rate can still be the right tool for a narrow job if the failures are understood. The number is useful only when it is paired with what the run produced.
What the number covers
On Apify, the catalogue uses the public 30-day run statistics where the platform exposes them. A successful run is a platform event. It says nothing about the accuracy of a price, the freshness of a listing or the number of empty cells in the dataset.
The sample matters. A tool with a handful of runs can move from 100 to 50 percent on one failure. A high-volume tool has a more stable rate but may also have more varied inputs.
Pair it with freshness
A maintained Actor with a recent build is more likely to match the current site than one that has not changed in a year. Freshness is not a guarantee either, but it is a useful second signal. The target page shows both the public rate and the update date.
Run a small completeness test
Pick a real input, keep the row limit small, and inspect the output. Measure the fields you need, not the fields the listing advertises. Save the result and date. That sample is more useful for your decision than any single platform metric.
Questions people ask
What is a good scraper success rate?
There is no universal threshold. A high rate on a large sample is a positive signal, but the schema, completeness and freshness still decide fit. A low rate on a tiny sample is not enough to reject a tool by itself.
Does a successful run mean the data is correct?
No. A run can complete while returning stale, sparse or misclassified records. Validate a sample against the source before relying on the dataset.
Why does a popular tool rank below a smaller one?
Adoption is one weighted input, not the whole formula. Field coverage, price, success rate and freshness can outweigh a larger user count. The target page shows the inputs.
How often should I recheck a tool?
When the source changes, when the schema changes, or before a production workflow depends on it. The catalogue records a check date so you can see how old the evidence is.
Sources
These sources support the general method on this page. A specific tool's price, schema or access route is on its target page with the check date.
Pick the website you need data from
Each website has a route map: every catalogue Actor for it, the fields each one returns, the price per 1,000 results and the official API beside them. Or search all 394 Actors by the field you need.