Same AI Question, Different Language: A 67,200-Response Audit
An audit of GPT, Claude and Gemini found that responses to the same Ukraine-war statement bank varied across 112 language and script conditions.
From this desk
Start with the most useful page
The featured page offers a strong introduction. The remaining cards help readers move through this editorial section without turning the archive into a service menu.
Read the flagship case study
How SearchEngineAnswer went from zero to early search visibility
This first-person build record connects publishing decisions with indexing evidence, Search Console observations, limitations, and the next measurement window.
Case study
Connect decisions with observed outcomes
Follow the site build, indexing evidence, changes, and limits instead of treating one visibility event as a universal rule.
Experiments
Change one material variable
Preserve the baseline, observation window, confounders, and decision rule so a result remains inspectable.
Evidence
Keep every claim close to support
Classify facts, observations, interpretations, recommendations, and opinions before drafting conclusions.
Latest research
Completed tests, case studies, and evidence reviews
Each research article identifies the question, evidence, method, result, limitation, and boundary on what can be inferred.
-
-
Algebraic Retrieval: 38 Tests Passed, With One Reproduction Limit
We ran the portable Algebraic Retrieval suite and verified its receipt hash. The full 11,429-document Java fixture remains outside this local test.
-
ChatGPT, Claude, Grok and DeepSeek Search the Web Differently
A 171,264-conversation study shows that four AI platforms differ at every search stage, from deciding to browse through selecting citations.
-
A GEO Measurement Paper Links AI Visibility to Business Impact, but Tests Simulated Answers
A new GMMM preprint connects generated-answer inclusion, question volume, system use and notice probability, but its empirical demonstration is simulated.
-
Chrome Now Measures Ad Count, Density and Weight in CrUX
Chrome added four experimental ad metrics to CrUX. We measured a fixed publisher panel and found why count, density, CPU and network weight need separate diagnoses.
-
What 30,000 Reddit Answers Reveal About Which Voices Survive AI Search
A 30,000-output study finds retrieval drives most variation while Reddit Answers favors early, top-level and more formal comments in its selected evidence.
-
Google Research Cut Retrieval Fan-Out Time 12× to 20×
Google researchers report that Retrieve-for-Train generated ten retrieval directions 12× to 20× faster in fashion and music benchmarks, while the diffusion model traded some recall for greater diversity.
-
Q2D-Web Shows Why AI Citations Depend on First-Stage Retrieval
A 190-million-document benchmark shows why retrieval recall, unique evidence coverage and downstream citations must be measured separately.
-
How to Measure the Noise Floor of an AI Visibility Score
A 1,200-observation research protocol for measuring how AI mentions, recommendations, and citations move when the monitored site does not change.
A useful standard
How to read SearchEngineAnswer research
A useful study makes its conditions visible. Treat each result as evidence within a declared sample and environment, not as a universal platform rule.
Method before conclusion
The question, inputs, environment, measures, and stopping rule should be understandable before the result is interpreted.
Failures stay in the record
Timeouts, absent results, rejected cases, and contradictory observations are data, not material to hide.
Limits remain visible
A measured association or one platform observation does not automatically establish causation or a permanent ranking rule.
References and next steps