Google AI Overviews & AI Mode
What AI Overviews are actually doing to your traffic
Impressions hold, clicks fall on informational queries, commercial queries are largely unaffected. That pattern has strategic consequences, and most of them argue for investing in the pages closest to a purchase.
The diagnosis
What is actually happening in the data
Most teams arrive at this page with a number — organic clicks down some percentage — and an assumption about the cause. The number is usually real and the assumption is usually only partly right.
- Clicks fell. Impressions did not.
- This is the signature of an answer surface rather than a ranking loss. You are still being shown; fewer people need to click. It is separable from an algorithm update, which typically shows as impressions falling, and from seasonality, which moves impressions and clicks together.
- The loss is concentrated, not general.
- Informational queries — how, what, why, definitions, comparisons of concepts — take the majority of the hit. Commercial and transactional queries are far less affected, because someone choosing a provider still needs to reach a site. Reporting a single site-wide percentage averages two very different situations together and produces panic where there should be a decision.
- The lost traffic was not worth what people assumed.
- When we weight the loss by conversion rate rather than volume, a large share of it turns out to have been traffic that never converted. That does not make it worthless — it had brand and authority value — but it changes the size of the problem considerably, and it changes what the right response costs.
- The instinctive fix makes it worse.
- The common reaction is to publish more informational content to recover the clicks. That competes directly with the surface that took them, using the format it is best at replacing. The productive move is the opposite: fewer, better informational pages written to be cited, and more weight on the commercial pages a generated answer cannot substitute for.
How we diagnose it
Segmenting the data before drawing a conclusion
We group queries by intent rather than by page — informational, commercial investigation, transactional, brand — and then look at impressions and click-through rate separately for each group over a long enough window to exclude seasonality.
That produces a picture almost nobody has before we start: which parts of the site are genuinely losing to answer surfaces, which are losing to competitors, which are simply seasonal, and which are unaffected. Those four situations need four different responses, and a single site-wide figure hides all of them.
We then weight the loss by what it was actually worth. Traffic that converted at close to zero and traffic that produced enquiries are not the same loss, and the budget response should reflect that rather than the raw click count.
Where the data genuinely cannot separate causes — and sometimes it cannot — we say so. An honest "we cannot isolate this" is more useful than a confident answer that sends a year of budget in the wrong direction.
The response
What we do once the picture is clear
- Rebalance the content portfolio toward queries that still reliably earn a click
- Consolidate informational pages that are now competing with each other for a shrinking click pool
- Rewrite the informational content worth keeping so it is genuinely worth citing
- Strengthen the commercial pages — comparisons, service pages, pricing education — that a generated answer cannot replace
- Establish a measurement baseline so the next shift is detected in weeks rather than quarters
- Report the effect segmented by intent, so leadership sees a decision rather than a scary number
Questions
What teams ask when clicks start falling
How do we know AI Overviews caused our traffic drop?
By the signature it leaves. AI Overview loss shows as falling click-through rate at broadly stable impressions, concentrated on informational query groups. An algorithm update usually shows as impression loss. Seasonality moves impressions and clicks together.
Segmenting Search Console by query intent separates these in most cases. Where it genuinely cannot be isolated, we say so rather than picking the explanation that suits a pitch.
Is SEO still worth investing in?
Yes, and in most businesses the case is stronger than before, though the allocation changes. Commercial and transactional queries — the ones closest to a purchase — have been far less affected, because someone comparing providers or ready to buy still wants to reach a site.
What has genuinely weakened is broad informational content published to capture top-of-funnel traffic. If that was the strategy, it needs rethinking. If the strategy was owning the searches that precede a purchase, very little has changed.
Should we publish more content to win the clicks back?
Usually not, and this is the most common expensive mistake we see. Publishing more informational content competes directly with the surface that took those clicks, and it is a contest you cannot win by volume.
The better response is to rebalance: fewer informational pages, better ones, written to be cited; and more investment in the commercial pages a generated answer cannot substitute for.
Can you get us cited in AI Overviews?
We can make you a more usable source and we cannot guarantee inclusion. The selection is not something anyone outside Google controls, and it varies by query and by user.
What we can do is make sure the pages worth citing are crawlable, clear, sourced and genuinely better than what is currently being drawn on.
Get an accurate read on what your traffic drop actually is
Give us Search Console access and we will segment it by intent before the call, and tell you how much of the loss is answer surfaces, how much is competitors, and how much of it mattered.
If we don't deliver the work we agreed to deliver for reasons within our control, you don't pay for the undelivered work. Read our guarantee
Last updated · Reviewed by Zubair Afzal