AI Overviews can change how people interact with search results, but a drop in clicks or lead quality does not identify the cause by itself. Check demand, query mix, site changes and conversions, then inspect relevant search results. Use the findings to improve the page and qualify enquiries rather than abandoning SEO.
The short version
- A traffic pattern is a finding to investigate, not proof of a penalty or AI effect.
- Measure enquiry quality with a clear definition and actual follow-up data.
- Use Search Console, analytics and customer feedback within their limits.
First, define what changed
“Traffic is down” can describe several problems. Fewer people may be searching, a page may have lost visibility, search results may have changed, or the mix of queries may be different.
Choose the affected pages and a comparable period. Record clicks, impressions, average position and useful conversions. Account for seasonality and any website, content or tracking changes.
Average position is an aggregate, so a steady number can hide changes in the queries receiving impressions. Look at relevant query groups before deciding what the overall figure means.
Can fewer clicks prove that AI Overviews caused the drop?
No. A page may receive fewer clicks when an answer appears above it, but the analytics pattern alone cannot establish that explanation.
Inspect a relevant question sample and record whether an AI Overview appeared in that observation. Keep the date, language and location context. Also check ordinary competitors, ads, other search features and changes in search demand.
Do not call a click drop a “signature” that proves one cause. A careful report separates the observed trend from the proposed explanation.
| Observed change | Checks to make | Conclusion to avoid |
|---|---|---|
| Clicks fall, impressions stable | Query mix, results features and demand | AI definitely caused it |
| Visibility falls | Indexation, access, competition and page changes | It must be a manual penalty |
| Direct visits rise | Known referrers and customer source answers | All direct traffic is AI |
| Enquiries fit poorly | Requested service, location and actual offer | All AI leads are low quality |
Is it a Google penalty?
A drop in clicks alone does not diagnose a penalty. Check Search Console's manual actions and relevant technical findings, then investigate the wider performance pattern.
Algorithmic changes, demand changes and page problems are different from a documented manual action. Avoid using the word penalty for every decline.
If several important pages suddenly become unavailable or excluded, investigate the access and indexation issues. If visibility remains but fewer people click, inspect the query and results context.
What does Google report about AI features?
Google's AI features guidance says its AI experiences sit within the overall Web performance reporting in Search Console. That does not let you attribute every organic visit or missed click to an AI Overview.
Use the reporting for the questions it can answer: which pages and queries changed and how the organic journey performs. Combine it with actual conversion information.
A screenshot can show that an AI answer appeared for one question. It cannot provide the total number of affected customers.
Are AI-related enquiries necessarily worse?
No. Quality needs a definition and evidence. Decide what counts as a suitable enquiry: the requested service, location, budget or other genuine fit criteria.
Compare those details with how the person found you where that information is available. If the source is unknown, keep it unknown.
An enquiry that arrived after an AI shortlist might be well informed or poorly matched. Your offer, public information and follow-up can affect the result. Do not generalise from a few messages.
Can analytics identify every AI visitor?
No. Some visits carry identifiable referrers and others do not. The available information depends on the tool, browser and journey.
Review known referral sources in analytics, but do not label all direct visits as AI traffic. Add a simple optional question about how the customer found the business and keep their wording.
Use that evidence alongside the landing page, service requested and eventual outcome. The tracking and attribution service explains the wider measurement work.
Improve the information buyers use to judge fit
If enquiries repeatedly ask for services you do not provide, clarify the offer. State the actual service coverage, process and important boundaries in visible text.
A UAE store should make delivery and seller policies easy to understand. A clinic should make the actual branch and available consultation clear. A B2B provider should explain its use cases and genuine capabilities.
Do not add arbitrary minimum budgets or qualification claims that the business has not agreed. The page should describe the real offer.
What should you change next?
Fix confirmed technical or factual problems first. Then improve the page's answer, supporting detail and internal links. Keep useful SEO work running while you review the evidence.
Judge the next period using comparable queries and business outcomes. If a guide attracts fewer visitors but generates suitable enquiries, that is different from a page losing both visibility and demand.
AdGrow's Google AI Overviews and AI Mode service connects these checks with useful website improvements. A free AI visibility audit can identify the starting priorities without promising a particular answer position.
Frequently asked questions
Can a click drop at a stable average position prove an AI Overview effect?
No. Inspect query mix, search demand, result features and your own site changes. The pattern can suggest questions to investigate, but it does not identify a cause by itself.
Are AI-generated enquiries always lower quality?
No. Define suitable enquiries and compare actual follow-up outcomes. Keep unknown sources unassigned and avoid generalising from a small number of messages.
Should we stop investing in SEO?
Do not make that decision from a traffic assumption. Keep useful technical and content work, then judge discoverability and business outcomes with evidence.
Can I identify every AI visitor from analytics?
No. Some visits carry an identifiable referrer and some do not. Customer source answers can add evidence, but direct traffic alone is not proof of AI discovery.
Sources and further checks
Platform guidance checked on 12 October 2026. Practical recommendations and hypothetical examples are AdGrow’s editorial guidance, not platform guarantees.