Search Parameter and Vendor Impact Research Note

Search Parameter and Vendor Impact Research Note

Finding

In September 2025, Google stopped reliably honoring the num=100 search-results parameter. Google did not publish a first-party announcement explaining a motive. Search Engine Land later reported Google's statement that the parameter was not formally supported.

Ahrefs provides the clearest first-party vendor record. Its September 18 update described reduced access to deep result pages and documented a staged restoration through October and November. The record proves that one major vendor had to change its monitoring system. It also shows why "the feature disappeared forever" is too broad: the URL shortcut stopped functioning as before, but Ahrefs later restored top-100 coverage through its own collection work.

Semrush's later AI-citation study treated the parameter change as a possible explanation for a correlation, then tested the data. That is a useful model for the episode package. A temporal match can justify investigation without establishing Google's purpose or a downstream causal chain.

Cost boundary

Replacing one response containing up to 100 results with ten responses containing up to ten results creates a ten-to-one request-count intuition. It does not prove a tenfold total cost for every operator. Caching, depth rules, retry behavior, infrastructure, access method, localization, blocking, and vendor architecture all change the result.

The public explainer may show the request arithmetic as an illustration. It must label total cost as vendor-specific.

Measurement boundary

The parameter was part of a results-collection interface used by third parties. It was not a ranking algorithm, a site change, or a Search Console control. Its removal could still influence reported Search Console impressions if automated result loads had previously generated impressions, but a site's own data must establish whether that happened.

Publication rule

State the observed change, the reported Google position, and the dated vendor response. Preserve Dalton Anderson's competitive interpretation as interpretation. Exclude legal conclusions about scraping and categorical claims about Google's motive.