AI Overviews Traffic Drop: What Small Businesses Do

Why is my traffic down when my rankings haven't changed?

Your rankings can hold while your clicks fall, because an AI Overview answers the question above your result. Fewer people need to click. The Pew Research Center tracked the browsing of 900 U.S. adults in March 2025 and compared visits to Google results pages with and without an AI summary. Users clicked a traditional result link in 8% of visits with a summary and 15% of visits without one. They clicked a link inside the summary in 1% of visits. They also ended their browsing session on 26% of pages with a summary against 16% without (Pew Research Center, July 22, 2025).

That is a real effect, but it is not proof that AI Overviews caused your drop. Traffic also falls for reasons that have nothing to do with AI: a core update, a spam update, a page that fell out of the index, a redesign that broke redirects, a season that ended, or a tracking tag that stopped firing. Most of the advice that ranks for this problem skips that step and goes straight to remedies. The next section is the step it skips.

How do I check whether AI Overviews caused the drop?

Compare clicks, impressions and position for the same pages and queries across two equal periods in Search Console. If clicks fell while impressions and average position held, AI Overviews are a likely cause. If impressions or position fell too, look elsewhere first. This takes about an hour for a small site.

1. Pick two equal periods. Open Search Console, then Performance, then Search results. Use the date filter's compare option: the last 3 months against the 3 months before, and also against the same 3 months a year ago if your business is seasonal.
2. Read the four totals. Look at clicks, impressions, click-through rate and average position. Write down the change in each.
3. Open the Queries tab and sort by clicks difference. Find the queries that lost the most clicks. For each one, check whether impressions and position held.
4. Open the Pages tab and do the same. Group the losers: informational pages (guides, definitions, "how to") or action pages (services, contact, pricing, booking).
5. Search your biggest losers yourself in a private window, from the country you sell in. See whether an AI Overview appears and whether your site is named in it. Results vary by place and over time, so treat it as a sample.
6. Check the Generative AI performance report in Search Console. It shows how often links to your pages were shown in AI Overviews and AI Mode, by page, country, date and device. Google says it was rolled out to all websites worldwide as of August 31, 2026, and it may not appear if your site has few such impressions. It reports impressions only, so it tells you where you are shown, not how many clicks that produced (Search Console Help).
7. Check the dates. Put the date your drop began next to Google's Search Status Dashboard.

Google counts traffic from AI features inside the ordinary Performance report under the Web search type, not in a separate bucket (Google Search Central). That is why the check above works by pattern: you cannot filter AI clicks out, so you read the shape of the drop instead.

Which symptom points to which cause?

Match what you saw in Search Console to the likeliest cause, then fix that one. The same traffic chart can come from five different problems.

Google's status dashboard listed these ranking updates in 2026, with start dates: a March 2026 spam update (March 24), a March 2026 core update (March 27), a May 2026 core update (May 21), a June 2026 spam update (June 24), an August 2026 spam update (August 18) and a September 2026 spam update (September 24). The page gives each a duration but no end date. If your drop starts inside one of those windows, suspect the update before the AI Overview.

How much do AI Overviews actually cut clicks?

Studies disagree on the size, and none measures your site. Two that state their method are worth knowing:

Read them carefully. Pew measured what people did, but for searches by U.S. adults in one month. Seer measured click-through on informational queries only, from its own clients, and its no-summary rate also fell over the period (from 2.74% in June 2024 to 1.62%), so the gap is not all down to AI Overviews. Seer also reported that in the third quarter of 2025, organic click-through was 0.70% when a brand was cited in the Overview and 0.52% when it was not. The author adds: "We cannot definitively prove that citation causes higher CTRs" (Seer Interactive). Pew also found that about 18% of the searches it analyzed produced an AI summary, so most searches in that sample did not.

A statistic you will see repeated, a "70% drop", is attributed to Seer in one widely syndicated article. We could not find that figure in Seer's own November 2025 update, so we do not use it. When a page quotes a percentage without naming a study, ask for the study.

Which of my pages are most at risk?

Pages that answer a question a summary can answer are at the most risk, and pages that lead to an action are at the least. The test is simple: could the answer be given in two sentences without the reader needing anything from you?

• Most exposed: definitions ("what is a 301 redirect"), short how-tos, lists of facts, and generic "tips" posts.
• Less exposed: pages where the reader must act. A service page, a booking page, a quote request, a local listing and a "near me" query all end in something a summary cannot do for them.
• In between: comparison and cost pages. A summary can quote a range, but a reader deciding whether to spend money often wants the table, the conditions and the source.

This is our reasoning, not a measured result for local businesses. We found no study that separates local service businesses from other sites, and the Forbes Australia article that ranks for this problem says only that local businesses had so far felt less impact than knowledge-driven ones. Test it on your own pages with the check above.

What should a small business do about it?

Fix the page type first, then your visibility, then your measurement. Google's own guide to generative AI search says optimizing for it is still SEO, and asks for content that "provides unique expert or experienced takes that go beyond common knowledge and the ordinary" (Google Search Central). Here is how that turns into decisions for each page type:

Three more points keep you out of trouble:

• Do not block your way out. To appear in AI features, Google says a page must be indexed and eligible to show with a snippet, with no extra technical requirements. A nosnippet tag or a blanket block to "stop AI" removes you from the answers that still name businesses. Our guide to getting your business recommended by ChatGPT and Gemini lists which settings remove you from which assistant.
• Do not buy a shortcut. Services that promise a place in AI answers are selling what nobody controls. Google says third-party tools do not have access to its internal ranking data. If you are weighing a quote, how much GEO costs lists the published prices.
• Do not mass-produce pages for every possible prompt. Google's guide says pages created for query variations primarily to manipulate rankings or AI responses violate its scaled content abuse policy.

If you have done the diagnosis and the cause is a ranking loss or a technical fault, that is ordinary SEO work, and it is what our search engine optimization service starts with: finding why pages lost position and fixing them in order of what is closest to bringing in an enquiry.

What if AI Overviews take the clicks that I used to get?

Then measure what you want from the page, not the click, and put more of the budget where a click still leads to an enquiry. That means looking at enquiries, calls, bookings and branded searches instead of sessions alone, since impressions in an AI answer can still build awareness of your name even when no one clicks.

Paid search is not exempt. Seer's data shows paid click-through on queries with an AI Overview falling from 19.70% in June 2024 to 6.34% in September 2025, against 19.1% to 13.04% on queries without one. Seer's summary: "Your cost-per-click may not have changed, but your efficiency likely cratered." That is from client accounts, so use it as a reason to check your own ads on informational terms, not as a forecast. Our guides to SEO against paid ads and what SEO costs a small business cover how to weigh the two.

Clicks that do arrive may be better ones. Google says clicks from results pages with AI Overviews are higher quality, but on the page we checked it gives no number. Test it on your own site: compare enquiries per visit on the affected pages before and after the drop.

What we can and cannot tell you

We can tell you how to read the data you already have. We cannot give you a percentage that applies to your site, and neither can anyone quoting one without your Search Console figures.

Our own site launched in October 2026, so it has no before-and-after history to show, and we make no claim about results for clients. What we do run every day is the same check described above: monitoring Search Console for pages that lose position or clicks, so we know whether a fall is a ranking problem or a click problem before we recommend work. We are an independent agency based in Islamabad, Pakistan, with clients in the US, UK, Australia and Pakistan, and we follow Google's spam policies: no bought links, no fake reviews, no mass-produced pages.