Search traffic is down. Check the ads before you blame AI
Did AI take our search traffic? It is a question worth settling before next year's budget is set, because the answer decides where the money goes.
New analysis from Kevin Indig gives a more careful answer than most. AI is part of the story. It is unlikely to be all of it, and the causes you can check this month deserve a look before anyone pays for an "AI search" fix.
What the numbers say
In Organic traffic is down, but it's not all AI, published in his Growth Memo newsletter on 26 October, Indig analysed traffic estimates from the data firm Similarweb for 10,000 websites in the US and 10,000 in France, covering the three years to August 2026.
In the US, visits from unpaid search results (the listings a business earns rather than buys) fell 13.2% between August 2024 and August 2026. Over the same two years, US visits from paid search, the ads Google shows among its results, rose 21%.
Across the sites studied, AI referral traffic made up roughly one visit in 200, about 0.5% of all visits. France saw a steeper fall in unpaid search (24.4%), which Indig puts mostly down to the different mix of sites in each sample.
Some caveats before anyone puts this in a deck. These are modelled estimates of visits, not people, sales or revenue, and the two samples don't represent the whole web. Indig also says plainly that the data cannot show what caused the drop.
AI is a suspect, and so are the ads
Indig is clear that AI plays a part. He says he has "no doubt" it reduces search traffic, and he thinks some of the loss comes from people asking chatbots to do jobs they used to do with a search.
His argument is that AI also takes the blame for damage other things are doing. In his opinion, ads now take up much more of Google's results page and are harder to tell apart from the unpaid listings, so their effect is underestimated and put down to AI. He cites research by Tom Capper at Stat, which found that ad blocks had nearly doubled in size across a sample of commercial search terms.
He also points to Google's own updates to how it ranks sites, including an unnamed one in January 2026 that he says wiped out rankings for some sites.
Ads don't explain everything either. By Indig's own sums, the growth in paid visits makes up only about 36.6% of the US loss in unpaid ones.
That is the point we would take into any client conversation. The causes you can check and act on come first: more ads above your listing, a Google update, how you compare with similar sites. The explanation that is fashionable to sell comes after, once the numbers point there. Get the order wrong and you spend money fixing a problem the client may not have.
Why the national figure misleads
The 13.2% headline hides very different outcomes. Publishers in the US sample lost about 41.5% of their unpaid search visits, and Indig links that to their product, content, competing directly with AI answers. Take publishers out and the US fall is 7.8%. The typical small site was roughly flat.
There are no UK figures in the study, so we wouldn't borrow its percentages for a British client. What travels is the method. Indig's own conclusion is that "comparable websites offer a more useful benchmark" than the country average. A retailer falling faster than similar retailers has a different problem from one whose whole category is falling.
For agencies, this changes what is worth selling. An "AI search" project sold to a client whose drop is really ads or a Google update won't move their numbers, and the agency owns that miss at the next review. A diagnosis that starts from a fair comparison is far easier to defend.
For in-house teams, it changes what to approve. If someone pitches you an AI search budget, ask first what your drop looks like next to sites like yours.
There is a fair counterpoint. Visits from AI referrals grew faster than any other channel in Indig's data, and that could continue. He also argues that what matters more than clicks from AI is whether your brand shows up when buyers research your category. So keep watching AI, and keep that visibility work on the list. Just don't build next year's budget on referrals that currently amount to about one visit in 200 on the sites studied.
What to do this week
- Benchmark against your peers. Ask your agency or analyst for your unpaid search trend next to three or four similar sites in your own market, of similar size and business, rather than against a national figure.
- Check what sits above you. Ask your search team which of your most valuable searches now show more ads above your listing, and whether the dates of your drop line up with a Google update.
- Give AI its own line. Report AI referral visits separately each month, so everyone can see how small or large they are and when that changes.
- Agree the diagnosis before the spend. Before any budget moves, agree in writing which explanation you are acting on, and what would change your mind.
We'd like to hear what you're finding on your own accounts. When search traffic has dropped, has the cause turned out to be AI, the ads, a Google update or something else?
Jules, founder of WYNStrategy Director helping agencies and marketing teams use AI properly. More about Jules