Traffic Value In The Artificial Intelligence Era
A page ranking first in Google used to mean a predictable share of clicks. With an AI Overview sitting above it, that share has been cut roughly in half, and for some queries it disappears almost entirely.
Every SEO conversation for the last two years has included some version of “AI is changing search.” What’s changed since is that the effect is now measured, repeatedly, by more than one independent source, and it’s larger than most marketing teams have adjusted their planning for.
The numbers, not the vibe
Ahrefs’ tracking of 300,000 keywords found that by December 2025, a page ranking first saw roughly 58% fewer clicks when an AI Overview appeared above it than an equivalent keyword without one, up from a 34.5% reduction measured in April 2025. Pew Research, working from a large sample of real search sessions, found a similar pattern: users click through to a traditional result about 8% of the time when an AI Overview is present, versus 15% without one. Digital Content Next’s tracking across 19 major publishers found a median year-over-year decline in Google referral traffic of 10% through mid-2025, with non-news publishers losing more (14%) than news publishers (7%).
These aren’t the same study reaching the same number by coincidence. Independent measurement from Ahrefs, Pew and Digital Content Next converging on the same direction and a similar rough magnitude is what makes this worth planning around now rather than waiting for more data.
It’s also worth being precise about what these numbers do and don’t say. A 58% click-through reduction on the top result doesn’t mean the query disappeared, it means the answer moved from a webpage to an AI-generated summary, often built partly from that same webpage’s content without crediting the visit back to it. The traffic didn’t vanish so much as get intercepted a step earlier in the journey, which is a subtly different problem than a simple decline in search volume, and one that content strategy alone doesn’t fully solve.
The decline isn’t evenly distributed
Two patterns are worth separating from the headline number. First, the query type matters enormously: informational queries with a clear, summarisable answer see the sharpest AI Overview presence and the steepest click declines, while queries requiring comparison, opinion, or a transaction see AI Overviews less often and retain more click value. Second, publisher category matters: Digital Content Next’s news-versus-non-news split suggests trust and recency, the things an AI summary is worse at replacing, correlate with which content categories are losing the least.
What still gets clicked
Content an AI Overview can’t fully substitute for still performs closer to historical norms: genuinely current information (an AI summary trained or retrieved on a lag is a weaker substitute for something that changed this week), first-person expertise and opinion (a summary can restate a fact, it can’t stand in for a named person’s judgement), and anything requiring interaction, a calculator, a tool, a comparison the reader actually needs to manipulate rather than just read.
That last category is worth dwelling on, because it points to a genuine, actionable response rather than just a description of the problem. A static article explaining how retention economics work is exactly the kind of content an AI Overview can summarise and intercept. A calculator that runs a reader’s own numbers can’t be meaningfully summarised the same way, because the value is in the interaction, not the explanation around it. That’s not a coincidence in how this site is built; it’s a direct response to the shift documented above.
Need to put a number on your next media decision?
Model the impact of a media or marketing decision on your own numbers, browse the full library of strategic calculators and decision tools, and get definitions straight on the industry terms that come up along the way, three free resources, ready whenever you need them.
What this means for where traffic value actually sits now
Organic search traffic as a channel isn’t dead, but its economics have shifted: the volume that converts into a visit is smaller, more concentrated in content an AI summary genuinely can’t replace, and less reliable as the primary acquisition channel it was treated as for the last decade. The practical implication is a bigger weighting on channels an AI answer can’t intermediate, owned email, direct, and first-party retention work, the same argument underlying Why Retention Stops At Loyalty: traffic you don’t have to keep re-earning through someone else’s algorithm is worth more now than it was two years ago, not because retention got more fashionable, but because the acquisition side of the ledger got structurally harder.
Free Playbook
The Retention Economics Playbook covers building the owned-channel and lifecycle case that matters more as organic acquisition value continues to erode.
Get the Retention Economics Playbook- Ahrefs, “Google says AI Overviews increase clicks…” AI Overview click-through rate update (December 2025 data)
- Pew Research Center, “Google users are less likely to click on links when an AI summary appears in the results” (22 July 2025)
- Digital Content Next, “Facts: Google’s push to AI hurts publisher traffic” (14 August 2025)
Traffic Value In The AI Era: FAQ
Is this decline the same across every industry?
No. Informational content with a clean, summarisable answer is hit hardest. Transactional, comparison-heavy, and genuinely time-sensitive content retains more of its historical click-through value.
Should we stop investing in SEO given these numbers?
No, but the expected return per unit of SEO investment needs to be recalibrated lower for informational content specifically, with more of the budget shifted toward content AI Overviews can’t easily substitute and toward owned channels less exposed to this shift.
Will this trend reverse if AI Overview accuracy or coverage changes?
Possibly at the margin, but the underlying behavioural shift, users getting comfortable not clicking through when an answer is presented directly, is unlikely to fully reverse even if AI Overview prevalence plateaus or declines slightly.
