AI search is no longer a sliver of traffic, and it does not reward the same pages Google's blue links reward. OpenAI put ChatGPT at 900 million weekly active users as of February 2026, and Similarweb's 2026 Generative AI Landscape report (via TechCrunch, July 27, 2026) found AI Overviews now show on 43% of Google searches, nearly triple the ~15% recorded a year earlier.
The bigger shift is where the citations come from. Ahrefs analyzed 863,000 keywords and 4 million AI Overview citation URLs and found only 38% of cited pages also rank in the traditional top 10, down from 76% just seven months earlier. Roughly two-thirds of AI citations now come from outside page one entirely. And when a business does get cited, it is not a vanity metric: Adobe's May 2026 retail data (1 trillion+ tracked visits) shows AI-referred traffic converting 54% better than non-AI traffic and growing 138% year over year.
Below: the full hub, grouped into four clusters (how big AI search actually is, where AI citations really come from, what getting cited is worth, and what to actually do about it), every figure attributed to a named source, with what we could not verify explicitly dropped rather than repeated.
How big is AI search, really, in 2026?
Start with scale, because it changes the stakes of everything that follows. AI answer engines stopped being a novelty and became a real, measured share of how people search, which means a business absent from those answers is now absent from a genuinely large audience, not a hypothetical future one.
ChatGPT has 900 million weekly active users
OpenAI announced on February 27, 2026 that ChatGPT surpassed 900 million weekly active users, alongside 50 million paying subscribers and more than 2.5 billion messages sent per day, roughly 29,000 messages every second. Many of those messages are exactly the buyer questions a local business wants to be named for: "who's the best [service] near me," "which tool should I use for X."
AI Overviews now appear in 43% of Google searches, up from 15%
Similarweb's 2026 Generative AI Landscape report, covered by TechCrunch on July 27, 2026, put AI Overview presence at 43% of tracked Google searches, up from roughly 15% about a year earlier, a 28-point jump. Neither Similarweb's public page nor TechCrunch's coverage discloses the exact geography or device mix behind the figure, so treat 43% as a large, directional panel estimate rather than a precise global rate. The trend is independently corroborated by Semrush Sensor, which separately tracks AI Overview presence near 48% of queries.
See how this plays out for a specific buyer question in Why AI Recommends Your Competitor, or check whether your own business shows up at all in Why Isn't My Business Showing Up in ChatGPT?
Where do AI citations actually come from?
This is the part that surprises most owners who already "do SEO." Ranking on page one of Google and getting cited inside an AI answer used to be roughly the same job. New data says they have split apart, and the gap is widening fast.
Only 38% of AI-cited pages also rank in Google's top 10, down from 76%
Ahrefs analyzed 863,000 keywords and 4 million AI Overview citation URLs and found only 38% of pages cited inside an AI Overview also ranked in the traditional top 10 organic results for that same query, a steep drop from 76% just seven months earlier. In plain terms: the page Google shows you first and the page Google's AI actually quotes are now, more often than not, two different pages.
Roughly two-thirds of AI citations come from outside page one entirely
The same Ahrefs dataset breaks down exactly where the other citations come from: 26.2% of cited URLs rank in positions 11 to 100, and 36.7% rank beyond position 100 or don't rank for that keyword at all, a combined ~63% from pages a user would never scroll to on Google itself. Google's AI now decomposes a single query into related sub-questions ("query fan-out") and pulls citations across every resulting result set, not just the one page optimized for the head keyword.
This is the exact mechanism behind why some businesses get named and others don't, covered in full in Why AI Recommends Your Competitor When Someone Asks for the Best in Town.
What is getting cited by AI actually worth?
Citations are a means, not the goal. The real question an owner should ask is whether a citation turns into a click, and whether that click turns into a customer. The newest retail data says yes, decisively, on both counts.
AI-referred traffic to retail sites grew 138% year over year
Adobe Analytics, tracking more than 1 trillion visits to U.S. retail sites, found AI-referred traffic (real click-throughs from AI platforms, not a proxy metric) grew 138% year over year as of May 2026, and 1,324%, more than 14x, since Adobe started tracking the category in October 2024.
AI-referred traffic now converts 54% better than non-AI traffic
The same Adobe dataset found AI-referred visitors converted at a rate 54% better than visitors from non-AI sources, a full reversal from a year earlier when AI traffic converted worse. Those visitors also spent 53% more time on-site and browsed 23% more pages, consistent with someone who already trusted the AI's answer and arrived pre-sold rather than starting a cold search from scratch.
Then vs. now: the one-year shift in four numbers
| Metric | ~1 year ago | Now (2026) | Source |
|---|---|---|---|
| ChatGPT weekly active users | ~400M (Feb 2025) | 900M (Feb 2026) | OpenAI / Search Engine Land |
| AI Overviews in Google searches | ~15% | 43% | Similarweb / TechCrunch |
| AI citations from Google's top 10 | 76% | 38% | Ahrefs / ALM Corp |
| AI-referred retail traffic (YoY growth) | baseline | +138% YoY | Adobe Analytics / Digital Commerce 360 |
Takeaway: every measured axis moved the same direction, fast. AI search got bigger, AI citations got less tied to traditional rank, and the traffic behind it got more valuable, not less. Nothing here is trending back toward "wait and see."
So what should a small business actually do about it?
None of this requires a rebuild. It requires writing (and structuring) pages the way AI answer engines actually read them: a real buyer question as the heading, the answer in the first two sentences, and a cited fact backing it up, repeated down the page.
Three things worth doing this month
- Rewrite your highest-traffic page's opening two sentences to directly answer the question a buyer would type. That is the chunk an AI Overview or ChatGPT actually lifts, per the Ahrefs data above, not the page as a whole.
- Put one real, named, dated stat under every major claim. AI answer engines favor pages with checkable evidence over pages with only marketing language, the same pattern behind every citation win logged on this blog.
- Track whether you're named when your own buyer questions get asked, not just your rank position. Rank and citation are now measurably different games, per the 38%-vs-76% shift above.
We won't invent a client result to make this feel more dramatic than it is: Neuron HQ has no paid case study to publish for this specific page yet. What we can show is our own site, built and structured exactly this way, and the two other posts on this blog that have already won a real, independently-confirmed citation from Perplexity (see the AI Inbox & Intake Agent and Best AI Receptionist for Dental Offices), proof this format works on our own account before we sell it on yours. See the full approach on Built by Neuron, or start with AI Marketing if you want this done for you, weekly.
Sources & methodology
This is a living hub. We review the figures on a recurring schedule, update the "last updated" date when we revise it, and replace any stat we can no longer attribute to a credible, named source. Where a study's own page returned a 403 or timed out on direct fetch (Ahrefs, Adobe), we cite the reputable outlet whose coverage we directly opened and verified instead, and we say so plainly rather than pretending we pulled the raw dataset ourselves. The load-bearing statistics on this page trace to:
- OpenAI: public announcement, reported by Search Engine Land, February 27, 2026. Source of the 900 million weekly active users figure.
- Similarweb: 2026 Generative AI Landscape report, covered by TechCrunch (July 27, 2026) and summarized by Digital Applied. Source of the 43%-of-searches / up-from-15% AI Overview figure; directionally corroborated by Semrush Sensor's independent ~48% tracking.
- Ahrefs: study of 863,000 keywords and 4 million AI Overview citation URLs, analyzed by ALM Corp (2026). Source of the 38%-vs-76% top-10 citation overlap and the 26.2% / 36.7% citation-position breakdown.
- Adobe Analytics: tracking of over 1 trillion visits to U.S. retail sites, reported by Digital Commerce 360 (data through May 2026). Source of the 138% YoY / 1,324%-since-Oct-2024 traffic growth and the 54% conversion, 53% time-on-site, and 23% pages-per-visit figures.
Last reviewed: August 2, 2026. If you cite this page, attribute the underlying study by name; we've kept the original sources visible so the chain is verifiable. Found a figure that's drifted? Email support@neuron-hq.com and we'll review it.
We'll show you exactly where you stand when the question gets asked.
Tell us your business and the questions you want to be found for. We'll check what ChatGPT, Perplexity, and Google's AI answers actually say today, and show you the specific gap between where you rank and where you get cited.
See the full approach on the AI Marketing page, browse more guides on the Neuron blog, or start from the Neuron HQ homepage. A real reply from the people who'll build it, usually within one business day.
Frequently asked questions
How many people now use ChatGPT or AI search every week?
OpenAI announced on February 27, 2026 that ChatGPT had surpassed 900 million weekly active users, alongside 50 million paying subscribers and more than 2.5 billion messages sent per day. That is roughly the population of Europe asking a chatbot questions every single week, many of them the exact buyer questions a local business wants to be named for.
How often do AI Overviews show up in a Google search now?
Similarweb's 2026 Generative AI Landscape report, covered by TechCrunch on July 27, 2026, found AI Overviews now appear in 43% of Google searches, up from about 15% a year earlier. Similarweb and TechCrunch do not publish the exact geography or device scope behind that number, but the direction is independently corroborated: Semrush Sensor separately tracks AI Overview presence near 48% of queries.
Do you have to rank #1 on Google to get cited by AI?
No, and it is getting less true every month. Ahrefs analyzed 863,000 keywords and 4 million AI Overview citation URLs and found only 38% of pages cited in an AI Overview also rank in the traditional top 10 for that query, down from 76% just seven months earlier. Most citations now come from outside the first page of Google results entirely.
Where do AI Overview citations actually come from, if not the top 10?
Per the same Ahrefs study, 26.2% of cited URLs rank in positions 11 to 100, and 36.7% rank beyond position 100 or do not rank for that keyword at all, roughly two-thirds of citations from pages a user would never see on page one. Google's AI decomposes a query into related sub-questions ("query fan-out") and pulls citations across all of the resulting result sets, not just the one page you optimized.
Does getting cited by AI actually bring real traffic?
Yes, and it is compounding fast. Adobe Analytics, tracking over 1 trillion visits to U.S. retail sites, found AI-referred traffic grew 138% year over year in May 2026, and 1,324% (more than 14x) since Adobe began tracking it in October 2024. This is measured referral traffic from AI platforms clicking through to a real site, not a proxy metric.
Does AI-referred traffic convert as well as normal traffic, or is it just curious clicks?
Adobe's May 2026 data found AI-referred traffic converted 54% better than non-AI traffic, a full reversal from a year earlier when AI traffic converted worse. Visitors from AI referrals also spent 53% more time on-site and browsed 23% more pages, consistent with someone who already got a trustworthy answer and arrived pre-sold rather than someone starting from zero.
So what should a small business actually do with these numbers?
Stop optimizing only for the traditional top 10 and start writing pages that answer a specific buyer question in the first two sentences with a real, sourced fact, because that is the chunk AI Overviews and ChatGPT actually lift. Track whether you get named when your own buyer questions get asked, not just rank position. Neuron HQ builds sites and content structured exactly this way; see how at AI Marketing.
The three ranking factors that decide who AI names when someone asks for the best in town.
The specific, checkable reasons a business goes unnamed, and how to fix each one.
Why a website's job changed from destination to source material, and what still matters.
The site itself is the demo: how Neuron structures its own pages to get cited.