AI Overviews Now Expand by Default: A Mobile SERP Audit

As of 30 September 2026, Google is testing AI Overviews that open at full length on some queries, without the old "Show more" button, which pushes the standard blue links further down the page. Google has not said how many searches are affected. The practical answer for site owners is to audit your own mobile results, because the effect depends on your queries, not on a global average.

Key Takeaways

• Reports from 28-31 August 2026 say Google is testing AI Overviews that expand automatically for some queries; Google has not disclosed scope, geography or query types.
• Your "position 1" can now sit several screens down on mobile, so rank alone no longer tells you what a page is worth.
• Advanced Web Ranking's July 2026 data shows position 1 CTR at 9.69% when an AI Overview is present versus 20.02% across all SERPs.
• You can audit this in about 45 minutes with Search Console, a phone and a spreadsheet, no paid tool needed.
• Pages that answer the query in the Overview still win when they offer something the summary cannot: a tool, original data, a decision aid.

What Google is actually testing

The story broke in late August. The Verge reported on 28 August 2026, and Search Engine Roundtable and Search Engine Land followed, that Google shows a fuller AI Overview for some queries, "pushing the standard list of web links farther down the page." Previously you saw a truncated answer with a "Show more" button. In the test, the button is gone and an "Ask anything" follow-up box loads already open.

Three details matter for planning:

• Google told Search Engine Roundtable the Overview "may dynamically expand for topics where our systems determine it's most useful for people."
• The expansion reportedly pauses if you have already started scrolling, so the layout does not jump under your thumb.
• Behaviour was described as inconsistent in early testing, and Google has not published the share of searches affected.

I am treating this as a live test, not a settled rollout. Test behaviour can be rolled back, widened or reshaped. What does not change is the direction: more of the first screen belongs to Google's answer.

What the click data says (and does not)

Numbers on this topic get abused, so here are the ones I trust, with their limits.

Read the third row carefully. Some datasets show CTR partially recovering from a trough, which is exactly why you should not extrapolate one headline. None of these studies measured the auto-expand test specifically; Google itself has not publicly quantified that. Anyone quoting "traffic will fall X% because of auto-expansion" is guessing.

Google's own position, given to Search Engine Roundtable by spokesperson Jennifer Kutz per the reports, is that users found the dynamic experience more helpful and engaged more deeply with follow-up exploration. Both things can be true: users may be happier and your click may still be worth less.

Why "we still rank #1" no longer settles it

I have seen this in client reporting. A page holds position 2 for six months, impressions are flat, and clicks slide. Rank tracking says everything is fine. The reason is not that the page got worse; it is that the page moved down the screen without moving down the ranking.

That is the argument for auditing by screen position, not just rank. Two useful measures:

• Scroll depth to first organic link: how many screen heights down does your result start on a phone?
• Overview presence and expansion: does the query trigger an Overview, and does it open fully?

The 45-minute mobile SERP audit

Step 1: Pick 25 queries (10 minutes)

In Search Console, open Performance, filter to the last 28 days, and sort by clicks. Take your top 25 queries by impressions where you average position 1-5. Export them. Add a column for query intent: informational, comparison, transactional, local.

Step 2: Check them on a real phone (20 minutes)

Use a phone, logged out or in a private tab, on mobile data or your normal Wi-Fi. Because the test is inconsistent, note that a single check is a sample, not proof. For each query record:

1. Does an AI Overview appear? (yes/no)
2. Does it open expanded, or with "Show more"?
3. How many screen heights until the first organic result?
4. Where does your listing actually appear, in screens?
5. Is your site cited inside the Overview?

Repeat the ten highest-value queries on a second day. Personalisation and test bucketing mean the same query can look different.

Step 3: Join with Search Console clicks (10 minutes)

Add columns for clicks and CTR from Search Console and sort by the queries where an expanded Overview appears. Compare their CTR against similar-position queries with no Overview. This is a directional comparison in your own data, which is far more useful than any industry average.

Step 4: Sort into three buckets (5 minutes)

• Protected: no Overview, or an Overview that does not expand. Keep doing what works.
• Exposed: expanded Overview, your page answers a simple factual question. This is where clicks are most at risk.
• Opportunity: expanded Overview that cites you. Being named in the answer is a different, more measurable outcome; see whether branded searches rise.

What Search Console can and cannot tell you

Google's Search Console generative-AI report was expanded worldwide on 31 August 2026 and shows impressions only, by page, country, device and date. That helps you see exposure but not clicks from those surfaces. Do not use it as a click report. For a fuller explanation of what that report shows, keep its limits in mind when you compare it with your own phone audit.

Also remember that Google's Search Console counts an impression when your result is on the page, whether or not a user scrolled to it. On a screen dominated by an expanded Overview, impressions can rise while real visibility falls. I would trust CTR movement on stable-rank pages more than impression counts.

What to change on the pages that are exposed

Give the click a reason

A summary can restate a definition. It struggles to replace a calculator, a template, a benchmark table drawn from your own data, a screenshot walkthrough or a worked example. For each exposed page ask: what does the reader get here that the Overview cannot give them?

Lead with the answer, then earn the scroll

Put a two or three sentence direct answer at the top. It improves your chance of being cited and it is what a human wants. Then make the second screen clearly more valuable than the first.

Rewrite the title and description for the click

If the summary answers the "what," your title should promise the "how" or the "which one." A title such as "What is a canonical tag?" invites the Overview to do the job. "Canonical tags: 7 mistakes I found in 40 audits" only works if the body delivers those items, so write it only where it is true. Google's documentation notes it may rewrite titles it finds unclear or inaccurate, so a title that matches the page content is safer than a dramatic one.

Measure brand demand, not only clicks

If an expanded Overview cites you, some readers will never click but will remember you. Track branded query impressions and direct traffic alongside organic clicks. It is imperfect, but ignoring it would be worse.

What I would not do

• Do not chase the layout by stuffing pages with FAQ blocks purely to be quoted. Cited answers still need substance.
• Do not decide on the basis of one phone check on one day.
• Do not cut informational content wholesale. Some of it now earns visibility instead of clicks, which is still worth something, and other pages will retain click value.
• Do not panic-buy a tool. The audit above answers the first question for free.

How this fits with the September spam update

Separately, Google began a spam update on 24 September 2026 with a rollout of up to two weeks. If your traffic moves in late September or early October, do not blame the Overview test automatically. Check dates: a drop that starts around 24 September with ranking movement points to a different cause than a CTR fall at stable rankings. Separating the two is the point of the audit, and the same discipline applies to any update.

A simple reporting template

For monthly client reports I now show three lines side by side: average position, CTR, and the share of tracked queries with an Overview. If position is flat, CTR falls and Overview share rises, the story is layout, not quality. If position falls with CTR, look at content and updates. That one chart prevents a lot of wrong conversations.

FAQ

Is Google making AI Overviews expand for everyone?

Not as far as has been reported. As of 30 September 2026 it is described as a test on certain queries, with inconsistent behaviour. Google has not disclosed the share of searches, countries or query types involved.

Does an expanded AI Overview hurt my rankings?

No. Your ranking position in the list of links is unchanged. What changes is how far down the screen that list starts, which can reduce clicks on the same position.

How much traffic will I lose?

Nobody can say for the auto-expand test specifically. Advanced Web Ranking's July 2026 data shows lower CTR when an Overview is present, but that is a general figure, not a measurement of this test. Your own Search Console data is the best predictor.

How do I check if my queries trigger expanded Overviews?

Search on a phone in a private tab and record whether an Overview appears, whether it opens fully, and how far you scroll to the first organic result. Repeat across two days because the behaviour is inconsistent.

Can I opt out of AI Overviews?

Google announced a generative-AI opt-out control on 31 August 2026, and I would read its details before using it since it is described as not a ranking signal. Blocking may reduce your visibility in those surfaces as well as your exposure, so decide per site rather than by reflex.

Does the Search Console AI report show clicks?

No, it shows impressions only by page, country, device and date, according to reports of the 31 August expansion. Use it for exposure, and your own performance data for clicks.

Should I stop writing informational content?

No, but be selective. Content that only restates a definition is most exposed. Content with original data, tools, or a decision framework still gives people a reason to click.

Is being cited in the Overview worth anything?

It can be. You gain brand exposure and sometimes clicks, but the value is harder to measure. Track branded search and direct visits alongside clicks and treat it as a supporting signal, not a replacement.

How often should I rerun the audit?

Monthly for your top 25 queries, and after any announced Search update. Since this is a test that can change quickly, a short recurring check beats a one-off deep dive.

Want help auditing your own SERPs?

I am Younus Fardeen, an organic growth and SEO strategist with 4+ years of marketing experience, working with edtech and startup teams on search, content and social. I ran organic growth on a project for Masai School (https://masaischool.com), where Instagram grew from 26K to 117K and LinkedIn from 50K to 200K. If you want a second pair of eyes on your mobile results, see my work and send a message through the contact form at https://younusfardeen.in.

Sources referenced: Search Engine Roundtable on CTR data, Advanced Web Ranking CTR study, Google title link documentation, Playwire on the August 2026 change, MLQ News on the test.