Index checker accuracy benchmark

Why this metric sits in the benchmark

Every indexation rate on this site is computed from index checks. If the checker were wrong often, the 95%+ and 85%+ figures would be wrong with it. Publishing the checker's own accuracy, about 96% on average, gives readers a measure of how much trust the other numbers can carry. The definition, verdicts that match the true state divided by all verdicts, is set out in the methodology.

How the checker decides

The IndexChex checker searches live Google results for each URL in four ways:

1. The bare URL.
2. The URL in double quotes.
3. site: followed by the URL.
4. inurl: followed by the URL.

If the URL appears for any of the four, it is marked indexed. Using several forms reduces false negatives, where a page is indexed but one particular query form fails to surface it. Checks cost 1 credit per URL, take up to 10,000 URLs per job and export to CSV.

Where checks go wrong

Errors on a public-results check come from a few recurring sources:

• Canonical consolidation. Google may index a different URL as canonical, for example a version with or without a trailing slash or tracking parameters. The submitted URL then does not appear even though its content does.
• Fresh indexation. A page indexed minutes ago may not yet surface for every query form.
• Operator behaviour. Google does not document site: as a complete index listing, so results for it can vary.
• Recently dropped pages. A page removed from the index can still appear briefly.

What is not published is how the 96% splits between false positives (called indexed, actually not) and false negatives (called not indexed, actually indexed), or how ground truth was established. Readers should not assume either direction dominates.

Effect on the indexation rates

A small error in the checker moves the measured indexation rate by a small amount. Wrong verdicts in both directions partly cancel out, so the net shift depends on how the errors are distributed rather than on their total.

Cost of verification

A scheduled re-check, set 1 to 5 days after a submission, costs the same 1 credit per URL and is reserved up front. Checking a standard batch at about day 5 adds one credit per URL to the submission credit, so verification roughly doubles the credit spend of a standard job while leaving instant jobs nearly unchanged (61 credits against 60).

Compared with other approaches

Google's URL Inspection tool is the authoritative view for a single URL, but only on a verified property, and it inspects one URL at a time. For third-party backlinks, public-results checks are the available option. No other indexer vendor checked for this site publishes a measured accuracy figure for its own checker. The checker in detail, with its query forms and their trade-offs, is covered on the network's checker accuracy page.

The measured tool is IndexChex, publisher of this observatory. The checker's place in the overall measurement is described on the benchmark page, and a plan for checking another vendor's results independently is on how to benchmark an indexer yourself.