Google AI Mode and Google AI Overviews share only 13.7% of their cited URLs and 18.7% of their cited domains on the same prompt, measured across 42,937 prompt days. AI Mode cites 12.95 sources per answer against 8.73 for AI Overviews, and named the tracked brand on 35.8% of prompt days against 27.3%. AI Mode also draws from a narrower pool, placing 48.9% of its citations in 100 domains where AI Overviews reaches 37.0%. There is no ranking position in AI Mode, so presence, order among named brands and the citation set are the only real metrics.
Google AI Mode tracking means running your buyers' questions in the AI Mode tab on a schedule and recording the answer, because AI Mode has no ranked results page to check and disagrees with the rest of Google about which sources to use.
That disagreement is the whole reason it needs separate tracking. On the same prompt on the same day, AI Mode and Google AI Overviews shared only 13.7% of their cited URLs across 42,937 prompt days. Reporting one Google figure averages two surfaces that barely overlap.
AI Mode is a separate tab in Google Search with a conversational thread, follow up questions and no ten blue links underneath. AI Overviews is the generated block inside the ordinary results page. Different entry points, different products, and Google says so in its Search Central documentation: "AI Mode and AI Overviews may use different models and techniques, so the set of responses and links they show will vary."
Both run the same retrieval pattern first. Google describes it as a query fan out technique, "issuing multiple related searches across subtopics and data sources", before assembling a reply. Neither surface is a ranked list, which is why a rank tracker tells you almost nothing about either.
Google sets one shared precondition worth knowing: "To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet."
Mention rate, per surface. Whether the brand was named at all. In our data AI Mode named the tracked brand on 35.8% of prompt days against 27.3% for AI Overviews, which makes AI Mode both the more generous surface and the one more often left unmeasured.
Citation slots and who fills them. AI Mode attaches 12.95 cited sources to an average answer, against 8.73 for AI Overviews and 4.76 for ChatGPT. Thirteen slots is a genuinely winnable competition, and the stored source list is what tells you which pages took them.
Domain concentration. This is where AI Mode surprises people. It puts 48.9% of its citations into a hundred domains, while AI Overviews reaches only 37.0% at the same cut. AI Mode gives each answer more slots while drawing them from a smaller regular cast, so entry is easier than persistence.
Single surface appearances. Of the prompt days where a brand appeared on Google at all, 51.2% appeared on only one of the two surfaces. Directionally: a brand named in an AI Overview also appeared in AI Mode 75.9% of the time, while a brand named in AI Mode also appeared in the Overview only 57.8% of the time.
The sub queries. AI Mode expands one prompt into several searches before answering, and about 95% of those sub queries have no traditional search volume, so a keyword tool cannot find them. Seeing them is usually the explanation when you hold one surface and lose the other, which is what query fan-out analysis exists to expose.
Position. There is no position in AI Mode. A tool reporting "rank 4" for an AI answer has invented a metric. What exists is presence, the order brands are named in, and the citation set.
A single check. Answers vary between runs. One screenshot is an anecdote; a rate across weeks is data. Any tool selling you a one off check is selling noise.
A combined Google score. Covered above, and it remains the most common reporting error in this category.
Traffic as a proxy for visibility. AI Mode sessions frequently arrive without a usable referrer, so analytics undercounts them badly. Visibility inside the answer is the reliable signal; traffic is a lagging and partial one.
The sister surface is easier to size. Across 970,061 tracked Google results pages in a 90 day window, an AI Overview was present on 71.0% of them, and where it appeared it sat above the organic results 99.5% of the time. Those blocks carried an average of 11.81 references against 8.84 organic results on the same page.
That 71.0% is far above published market rates, and the reason is the sample rather than a contradiction. Ahrefs, working from 55.8 million AI Overviews across a 590 million keyword index, found AI Overviews on 9.46% of desktop keywords and roughly 16% in the United States. Their denominator is an entire keyword index including navigational terms. Ours is prompts customers chose because they describe a buying decision, which is exactly the language that triggers a generated block.
The useful reading is that AI Overview exposure depends almost entirely on which questions you ask about. A category benchmark beats a global average every time.
1. Split the two Google surfaces in your prompt library from day one. Retrofitting the split after six months of blended history loses the comparison.
2. Run daily and report a rate. Weekly is the floor. Anything less and normal variance looks like a trend.
3. Store the cited URLs, not just the count. The list of pages that won the answers you lost is the only output that produces a next action.
4. Track the same prompts on at least one non Google engine. Otherwise you cannot tell a Google change from a market change.
5. Check crawlability before content. Most AI crawlers do not execute JavaScript, and in an audit of 20,293 pages across 524 sites, 8.9% returned nothing a text first crawler could read. A page that returns an empty document cannot be cited by either surface, which is what AI crawler and bot traffic analysis is for.
The honest answer is that the evidence is thinner than the advice usually implies.
The Princeton GEO study tested nine content strategies across 10,000 queries. Adding cited sources, direct quotations and statistics produced 30% to 40% relative improvement on the metric it measured. Authoritative tone produced no significant improvement, and keyword stuffing landed at or below baseline. The lifts are also rank dependent: in the study's own table, adding sources moved a rank five page by 115.1% and moved the rank one page down by 30.3%. It is a redistribution effect, not a promise.
A separate 2026 analysis traced 62.2% of citation failures to semantic alignment, which in plain terms means the page answered a slightly different question than the one asked. That is the cheapest thing to fix and the least glamorous, and it is what AI content optimization is aimed at.
One requirement filters this list faster than any feature comparison: the tool has to report AI Mode and AI Overviews as two separate lines, and it has to cover both on the plan you buy.
Check the entry tier, not the marketing page. Profound's Starter plan at $99 a month billed yearly tracks ChatGPT only, so it cannot see either Google surface. ZipTie covers Google AI Overviews but not AI Mode, on every plan, with no add-ons. Peec AI and the Semrush AI Visibility Toolkit let you pick three models of your choice, which means AI Mode competes with ChatGPT and Gemini for a slot.
Finseo is the pick here because both Google surfaces are on every plan, alongside thirteen other engines, and because it stores the cited source list per answer rather than a count. With 13.7% URL overlap between the two surfaces, that stored list is the only way to see which pages took the slots on the surface you lost. Rankscale also covers both surfaces across its tiers and is a reasonable alternative where attribution is not part of the requirement.
If you need this on a small budget, Otterly.ai includes Google AI Overviews from $29 a month, with Google AI Mode as a paid add-on starting at $9 a month on the Lite plan. Price the add-on before assuming it is covered.
Figures attributed to Finseo come from tracking data aggregated across accounts, with no customer, project or private domain identifiable. The citation overlap covers 42,937 prompt days in a 14 day window and includes only prompt days where both surfaces returned at least one citation. Brand co-occurrence covers 45,024 prompt days in a 14 day window and drops any prompt day where one surface returned no answer. AI Overview placement covers 970,061 tracked results pages over 90 days. The crawlability audit covers 20,293 URLs across 524 domains.
Prompt sets are chosen by the customers who run them, so the corpus is large but not random, and the 71.0% presence figure in particular should not be read as a market rate.
How do I track Google AI Mode? Run your prompts in AI Mode on a schedule, record the answer text and the cited sources, and report a mention rate over time rather than a single check. Track AI Overviews separately.
Is AI Mode the same as AI Overviews? No. Google states the two may use different models and techniques, and we measure them sharing only 13.7% of cited URLs on the same prompt.
Can I see AI Mode in Search Console? Not as its own surface with reliable separation. Search Console does not break AI Mode out in a way that supports per surface reporting, which is why dedicated tracking exists.
How many sources does AI Mode cite? An average of 12.95 per answer in our data, with 94.6% of answers carrying at least one citation.
Why do I appear in AI Overviews but not AI Mode? Because the two surfaces retrieve differently. Just over half of all brand appearances on Google in our data existed on only one of the two.
Ours
External
Disclosure: the tracking figures in this article come from Finseo, an AI visibility platform. External figures are attributed to their original source.
