Performance Runner

How often AI models recommended each domain specifically via web-search fanout queries when answering Performance Runner queries, a subset of the runs behind the main LLM Search Visibility report, not the full combined dataset. The left chart shows raw recommendation counts, the right shows what percentage of queries included that domain.

Score (below) is the primary ranking metric: a weighted-reciprocal-rank sum across every model and run, where each model's contribution is scaled by that model's influence weight.

Only these five models ran web-search fanout queries. The checkboxes filter these charts and the Citations / Citation Sources lists below; updates automatically as you check/uncheck.

Every URL the models actually cited in their final answers to Performance Runner fanout queries, aggregated by how often each URL was cited. Filtered by the model checkboxes above.

Every URL the models' web searches retrieved while answering: the wider pool the citations above were drawn from, aggregated the same way and filtered by the same model checkboxes.

Pick a model for each column to compare their recommendation patterns side by side.

Every domain the AI models recommended in web-search fanout runs for Performance Runner, ranked by the influence-weighted score from the Score chart above, with each domain's link authority from the Common Crawl web graph. Click a domain for its site profile.