Not yet, not directly. Google's Search Console shows how many times a page appeared in AI Mode, never which company the answer named; Reflexa's own Index already records that for ChatGPT, Claude and Perplexity, on the same buyer questions. Until Google publishes answer-level data, any "AI Mode vs ChatGPT" comparison is one-sided.
Not with hard numbers, because Google doesn't publish which companies AI Mode's answers name, only how many times a page was used as a source. The new AI performance report in Search Console groups AI Mode together with AI Overviews and Discover's AI features into one impression count, with no query-level breakdown, covered in full in Google Search Console's new AI performance report. Reflexa's own AI search statistics tracking currently covers ChatGPT, Claude and Perplexity, the three engines where a buyer question and its named answer can both be recorded. Google AI Mode is on our roadmap, but only once there's a comparable way to see the actual names in an AI Mode answer, not just an impression count.
Do this: treat any claim that "AI Mode recommends X more than ChatGPT" as unverifiable today. Neither Google nor any third party publishes AI Mode's answer text at scale.
A count of impressions, and nothing about what the answer said. Google finished rolling the report out to every Search Console property worldwide on August 31, 2026, after announcing it in June 2026. It tells you a page was pulled into an AI Mode answer somewhere, for some query, in some country; it does not say which query, whether the answer named your business, or whether it named a competitor instead.
Proof: Google Search Central blog, "Introducing Search generative AI performance reports in Search Console" (June 2026); Search Engine Land, "Google Search Console AI performance reports and Search generative AI control rolling out globally" (August 2026).
Bottom line: an AI Mode impression and an AI Mode recommendation are different events, and only the first one shows up in Search Console. See the gap between an impression and a recommendation for the full breakdown.
That they don't always agree, even on an identical buyer question. Each engine sources and ranks its answer differently: ChatGPT blends training data with a live search pass, Claude attaches an inline citation to every claim it pulls from the web, and Perplexity shows a visible sources panel a reader can check directly, detailed in ChatGPT vs Claude vs Perplexity. That's why the Reflexa AI Visibility Index runs every buyer question through all three engines, three times each, and records the result per engine in its evidence log rather than blending them into one number.
| Engine | Named companies tracked today | Source |
|---|---|---|
| ChatGPT | Yes, per buyer question, 3 runs | Reflexa AI Visibility Index |
| Claude | Yes, per buyer question, 3 runs | Reflexa AI Visibility Index |
| Perplexity | Yes, per buyer question, 3 runs | Reflexa AI Visibility Index |
| Google AI Mode | No, impressions only | Search Console AI performance report |
Rule of thumb: a business can pass one engine's bar and fail another's on the exact same buying question, which is a reason to check more than one, not a reason to trust whichever one happens to name you.
Because the two surfaces pull from different sources, even when the search box looks the same. Ahrefs checked 15,000 prompts and found only 8-12% of the links cited by ChatGPT, Gemini and Copilot also appear in Google's top 10 organic results for the same prompt. It's a strong signal that a Google-only strategy leaves the AI surfaces, AI Mode included, mostly uncovered. The full context is in Ranking #1 on Google no longer means AI recommends you.
Proof: Ahrefs, AI search citation overlap study (2026).
Because AI answers are non-deterministic and every engine has its own sourcing logic, so one lucky mention on one engine is a snapshot, not a measurement. The Reflexa Index re-runs every question three times per engine for exactly this reason, and the same logic applies to Google AI Mode once it becomes measurable: a single well-timed impression would tell you as little as a single ChatGPT answer does today. Weekly re-checks catch the drift a one-off answer misses, covered in how often to re-check your AI visibility.
Do this: whatever engine you check, ask the same buyer question at least three times before drawing a conclusion about who it names.
None of the steps below need a purchase, and most take a few minutes each:
Bottom line: you can already measure who ChatGPT, Claude and Perplexity name; treat any AI Mode comparison as unmeasured until Google publishes more than an impression count.
Sources: Google Search Central blog, generative AI performance reports (June 2026) · Search Engine Land, AI performance reports rolling out globally (August 2026) · Ahrefs, AI search citation overlap study (2026) · Reflexa AI Visibility Index, September 2026
Not with public tools today. Search Console's AI performance report shows impression counts by page, country and device, but not the query asked or the names AI Mode's answer gave, and Google has not published an answer-level export of AI Mode results.
Not yet. Reflexa's AI Visibility Index currently runs buyer questions through ChatGPT, Claude and Perplexity, three times each, and records exactly which companies each one names. Google AI Mode is on the roadmap once Google exposes comparable answer-level data.
Not always. Each engine sources and ranks answers differently, so a business can clear one engine's bar and miss another's on an identical buyer question, which is why the Reflexa Index evidence log tracks each engine separately.
Because the two surfaces draw on different sources. Ahrefs found only 8-12% of links cited by ChatGPT, Gemini and Copilot also appear in Google's top 10 for the same prompt, across 15,000 prompts.
Ask your real buyer questions to ChatGPT, Claude and Perplexity yourself, or run the free check at dashboard.reflexatechnologies.com/audit, which does it across all three in about three minutes and shows the answer text and sources.
The free check runs real buyer questions across ChatGPT, Claude and Perplexity — 3 minutes, evidence included.