Methodology note: retrieval routes
Change the searcher and the sources change. The recommendations don't.
When an AI assistant answers a shopping question, how much of the answer is the model, and how much is the search engine feeding it? We measured it: identical prompts, the same models, two different retrieval routes, called seconds apart.
- brand overlap between routes: who AI recommends barely moves
- 0.83-0.92
- source-domain overlap between routes: what AI reads almost completely changes
- 0.09-0.27
Measured June 2026 on 20 frozen streaming prompts, ChatGPT, Claude and Gemini, paired calls in the same minute. Directional, not definitive.
Contents
Same prompt, same model, same minute, two search layers
Twenty frozen streaming prompts from our monthly study, run through each model twice in paired calls: once through the single shared retrieval layer our research uses (the Perplexity Agent API), and once through each vendor's own API with its native web search (OpenAI's search tool, Anthropic's web search, Google Search grounding). Same prompt, same minute, so day-to-day drift can't masquerade as route divergence.
Finding 1
The recommendation barely moves when the search engine changes
The set of brands each model named overlapped 0.83 to 0.92 (Jaccard) across routes. Who AI recommends is model behavior. It survives a complete change of search engine.
Finding 2
The sources behind the answer almost completely change
The cited domains overlapped just 0.09 to 0.27 across routes. The same model, asked the same question, reads an almost entirely different web depending on who does the searching. Even grounding behavior is route-dependent: ChatGPT searched the live web twice as often on its native route; Claude searched less.
| Engine | Brand overlap | Source-domain overlap | Grounded (shared) | Grounded (native) | YouTube / Reddit share, native |
|---|---|---|---|---|---|
| ChatGPT | 0.92 | 0.09 | 25% | 55% | 0% / 0% |
| Claude | 0.83 | 0.27 | 95% | 60% | 1% / 0% |
| Gemini | 0.83 | 0.15 | 95% | 100% | 6% / 6% |
Overlap is Jaccard similarity between the two routes across the run (1.0 = identical sets). *Gemini's native leg ran on the stable flash model; its shared-route leg uses the preview flash model, so its row carries a model-version caveat the others don't.
What this means for brands. You cannot change what a model believes this quarter, but the sources are an open competition that restarts with every searcher. The same answer is being assembled from different shelves, and the shelves are where you can show up.
The routes disagree most on creator and community content
The shared retrieval layer cites YouTube and Reddit on 5 to 12 percent of citations. Native Google grounding also cites both, at about 6 percent each. Native OpenAI and Anthropic search cite essentially neither. If your category's AI answers are shaped by creator videos and community threads, which engine a buyer asks, and what searches behind it, decides whether that content reaches them at all.
Why our research holds retrieval constant across all four models
It is the only way a cross-model comparison isolates the models themselves. This measurement is the receipt for that design choice, and for its boundary. Recommendation findings in our reports are model-attributable and hold across routes. Source-mix findings describe the supply the shared layer feeds the models, and every report says so.
What this means for brands. Read a mapou brand ranking as a statement about the model. Read a source mix as a statement about the search layer behind it.
The streaming study these prompts came fromHow we measured this
- Prompts
- 20 frozen streaming prompts from the monthly study. One category, one day: directional, not definitive.
- Routes
- Shared route: the Perplexity Agent API used for all mapou research. Native route: each vendor's own API with its native web search (OpenAI's search tool, Anthropic's web search, Google Search grounding).
- Pairing
- Each prompt ran on both routes in the same minute, so day-to-day drift can't masquerade as route divergence.
- Overlap
- Jaccard similarity between the two routes' sets across the run (1.0 = identical sets), for named brands and for cited domains.
- Caveats
- Gemini's native leg ran on the stable flash model; its shared-route leg uses the preview flash model, so its row carries a model-version caveat the others don't. Grok's native search API was mid-migration at measurement time and is excluded.
Measure your brand on this instrument
mapou measures how AI assistants alter commercial decisions inside high-intent consumer categories. Methodology notes like this one document the instrument itself.