AI Overviews and website clicks: what this study can tell you.

Check the business effect before changing the budget

A report about AI Overviews and click-through rate should help you decide what to do with your website. For a training provider, that means understanding what happened to course enquiries and enrolments. A recruitment owner needs to know about employer conversations and briefs.

My recommendation is to start with those results, then investigate the pages behind them. A traffic decline deserves attention, but the explanation needs more than a screenshot of an AI answer.

Multiply reviewed client search data to see if it could separate the effect of AI summaries. The review confirmed that summaries appeared on many of the selected searches. It could not put a reliable number on the visits gained or lost because of them.

AI summaries appeared on most of the searches checked

On one Australian training provider’s account, 221 of 243 selected searches showed an AI Overview when checked in August 2026. That is about nine in ten. The searches met a repeat-appearance threshold across the two periods described below; they represent a selected part of one provider’s search activity.

The result gives that provider a reason to inspect how its pages appear in search. It should not become a forecast for another business, or a claim about every Google search.

Source: Multiply Labs’ retained search-result classifications, regenerated 28 August 2026. Australian searches only. Selection used Google Search Console records for 13 July–9 August in 2025 and 2026.

The review could not tell how many visits AI changed

Only 22 selected searches had no AI Overview. Once searches were grouped by their position in Google, there were too few comparable examples to draw a dependable conclusion about clicks.

The timing also limits the result. The team checked for AI summaries after the period covered by the click figures. That check does not establish which summaries a person saw when they searched during the earlier period.

These limits leave the effect unresolved. The review supports a closer look at the provider’s search results, not a percentage reduction to apply to its traffic or budget.

Ask your marketing team to connect the report to enquiries

Ask for a short review of the pages that changed most. It should answer these questions:

• Which pages lost visitors? Separate general advice articles, course or service pages, and searches for your business by name.
• What happened to qualified enquiries? Check the relevant forms and calls, and confirm the tracking still works. For a training provider, follow the applications through to enrolments where records allow.
• What else changed? Ask about the page’s position in Google, seasonal demand, and recent website changes before attributing the decline to AI.
• What does a prospective buyer see? Review a few important searches together. Check the information about your business, the pages Google links to, and the next step offered on your website.

Use the answers to choose the next piece of work. A course page with unclear fees needs a content fix. A broken enquiry form needs repair. A useful article that has lost its search position needs a different investigation.

For content, Google’s guidance for AI features applies the usual search principles: useful pages, accessible content, and accurate information. Google specifies no special additional requirements for AI Overviews. A fixed answer length or a particular table format cannot promise inclusion.

The answer engine optimisation guide explains the topic in more detail. The AI SEO guide covers the wider changes, and Multiply’s AI SEO and AEO service covers work on your own site.

How the review was done

The team used Google Search Console, Google’s report of a site’s search performance. It selected searches that appeared at least 30 times in each of two 28-day periods: 13 July–9 August 2025 and the same dates in 2026. It filtered the Australian account to Australian searches and the UK account to UK searches.

The selection excluded searches containing the configured business-name terms. That filter was imperfect: a person’s name remained in the Australian sample. The team then checked the selected searches for an AI Overview and compared click-through rates within groups of similar search positions.

The saved classification file supports the count reported above. The limited comparison group, later classification date, and imperfect filtering restrict what can be concluded about clicks.

Frequently asked questions

How do I optimise for AI Overviews?

Start with useful, accurate pages that answer your buyers’ questions. Google says its usual search guidance applies to AI Overviews, with no special extra requirements. A clear answer helps the reader; it does not guarantee that Google will feature your page.

Did this study show that AI Overviews reduced clicks?

The review could not establish that. Most of the selected searches showed an AI Overview, so there were few searches left for comparison. The summaries were also checked after the period covered by the click figures.

What is a good Google click-through rate?

Click-through rate is the share of times your result appeared that led to a click. Your position in the results and the type of search affect that figure. Ask your marketing team to compare similar pages and searches over time, then show what happened to enquiries. A site-wide average needs that context.

Should I reduce my content budget if visits fall?

Review the pages responsible for the decline first. A fall in visits to a general advice article calls for a different response to a fall in enquiries through a course or service page. Decide which work to continue after checking the enquiries and sales that matter to your business.

Sources

Multiply Labs’ August 2026 review used client Google Search Console records and retained search-result classifications. Client-identifying search terms remain private. Google Search Central’s AI features and your website supports the guidance on content and inclusion requirements; checked 10 September 2026.