Contents
TL;DRAI has changed SEO in two ways. Answers now sit above the links, so fewer people click, and the work itself, the research, analysis and drafting, can be done far faster. What AI still cannot do is judge strategy and quality. For businesses paying an agency $3,500 to $5,000 a month, that is the real shift.
How has AI changed SEO?
AI has changed SEO in two ways at once. First, search results now often start with an AI-written answer, so fewer people click through to websites. Second, the research, analysis and drafting behind SEO can be done much faster, which changes what an agency's time should cost and how long a change should take.
Neither change makes SEO obsolete. Google says optimising for its AI features is still SEO. But both change where you need to be found, and what you should be paying for.
8% against 15%
click-through, with an AI summary against without
Pew Research Center, July 2025
7.3% to 1.6%
position-one click rate on AI Overview queries over two years
Ahrefs, February 2026
61%
fall in organic clicks on AI Overview queries
Seer Interactive, November 2025
68%
of US searches ended with no click, 60% in 2024
SparkToro, June 2026
Are fewer people clicking because of AI answers?
Fewer people do click when an AI answer appears, in every study we found, though the size depends on who measured. Pew Research watched 900 US adults browse: they clicked a normal result in 8% of visits that showed an AI summary, against 15% without one. Studies run by companies that sell SEO tools found larger drops.
| Study | What it found | Method and who ran it |
|---|---|---|
| Pew Research Center (July 2025) | Clicks: 8% with an AI summary, 15% without, 1% on its link | 900 US adults in a browsing panel, March 2025; non-commercial |
| Ahrefs (February 2026) | Position-one clicks on AI Overview queries fell from 7.3% to 1.6% | 300,000 keywords in Search Console data; vendor, and the comparison is modelled |
| Seer Interactive (November 2025) | Organic clicks on AI Overview queries fell about 61% | 3,119 informational queries, 42 client organisations; agency |
| SparkToro (June 2026) | 68% of US Google searches had no click (60% in 2024) | Similarweb clickstream data; vendor, and the data sets differ between years |
| Semrush (December 2025) | Queries showing an AI Overview: 6.5%, then 24.6%, then 15.7% | More than 10 million keywords; vendor |
Some detail the table leaves out: the Pew figure of 1% is for clicks on a link inside the AI summary, Ahrefs measured its fall over two years, Seer Interactive over 15 months, SparkToro's 68% covers January to April 2026 and the Semrush shares are for 2025. Google's own view differs. In August 2025 it said overall click volume was "relatively stable" and the clicks that remain are higher quality, without publishing figures. We could not verify either side fully, so read the direction, not the decimals.
Do people use AI to search?
People do use AI to search, and more every year. Pew Research found 49% of US adults used AI chatbots in 2026, up from about a third in 2024, and 42% used them to look things up. OpenAI's chief executive said in October 2025 that more than 800 million people used ChatGPT every week, an executive's figure rather than an audited one.
Gartner predicted in 2024 that traditional search volume would fall 25% by 2026. We could not find any measurement confirming or refuting that, and Alphabet reported Search revenue up 17% in the quarter to June 2026. Treat the forecast as a forecast.
What does AI do well in SEO?
AI does research, analysis and speed well in SEO. It can read a site and a data export in minutes, spot ranking drops, group keywords, draft a first version of a page and write the fix list that an analyst used to spend a week on. Google agrees it can help with researching a topic and adding structure to original content.
It also removes waiting. A system plugged into live data can notice a change and flag it the same day, instead of in next month's report. That is where most of the time in an agency used to go.
What does AI still do badly?
AI still does judgement, accuracy and originality badly in SEO. The Tow Center at Columbia tested eight AI search tools on identifying news articles and found them collectively wrong on more than 60% of queries, and a European Broadcasting Union and BBC study of over 3,000 answers found 45% had at least one significant issue.
60%+
of news queries answered wrongly by eight AI search tools
Tow Center, Columbia Journalism Review, March 2025
45%
of 3,000+ AI news answers had a significant issue
EBU and BBC research, summarised by AIB, October 2025
49%
of US adults use AI chatbots, 42% to search
Pew Research Center, June 2026
- Strategy. Which customers to chase, which pages matter, and what to leave alone are business decisions.
- Quality. Google says generative AI used to mass-produce pages with little value can break its scaled content abuse policy, and that you should fact-check what it produces.
- Originality. Google says non-commodity content, with a unique point of view from first-hand experience, will likely matter more over time than any other tip.
- Consistency. SparkToro found that asking an AI the same question for brand recommendations gave the same list in fewer than 1 in 100 runs.
That is why we say "No AI slop". AI does the heavy lifting, and an experienced human checks every word before it goes live.
What does this mean if you pay an agency?
If you pay an agency, the question is what AI changed on your invoice. If your agency still charges $3,500 to $5,000 a month for a team of five and takes weeks to make a change, the AI savings are going to the agency, not to you. A better question than "do you use AI?" is "how fast do changes go live, and who checks them?"
| AI does well | A person still has to | |
|---|---|---|
| Research and analysis | Read a site and data export in minutes, group keywords, spot drops | Decide which findings matter for the business |
| Drafting | Produce a first version of a page or a fix list fast | Check every fact and word, and add first-hand experience |
| Speed | Flag a change the same day instead of in next month's report | Approve what goes live |
| Accuracy and originality | Little: the tests above found large error rates | Own the quality, as Google's policies on scaled content expect |
- How many people touch my website when I ask for a change?More than two is a warning sign.
- How long does a simple change take?Hours is possible now. Weeks is a process problem.
- Who checks the AI's work?A named person with experience, not nobody.
- Do you cover AI search as well as Google?Being named in answers matters as well as ranking in lists.
- Can I leave and keep my website?You should always be able to.
Where Digital Monday fits
Full disclosure: this is our blog. Digital Monday is $500 a month. A small team and our engine do your SEO and AEO, changes go live within 24 hours, you get a plain report of what changed, and you can leave anytime and keep your website. Our guarantee: Results in 60 days, or your first two months back. No questions asked.
The scope of a $500 service is different from a traditional $3,500 retainer, and we say so. We are built for business websites that bring in enquiries, not online stores or custom web apps. See how our SEO works, what AEO is or what SEO costs.
Questions people ask
01Is SEO dead because of AI?
No. Google says optimising for its AI features is still SEO. But answers now sit above the links, so fewer people click, and being named in the answer matters more.
02Does AI-written content hurt my rankings?
Not by itself. Google says AI content is not banned, but mass-produced pages with little value can break its spam policy, so a person should check and improve what AI drafts.
03Can AI replace my SEO agency?
It can replace much of the research and drafting time. It cannot replace strategy and quality judgement. The question is how much of your agency's price still pays for the waiting.
04How many people use AI to search?
Pew Research found 49% of US adults used AI chatbots in 2026, and 42% used them for information searching.
05What is AEO?
Answer engine optimisation: getting your business named in AI answers as well as search results. What is AEO?
Sources and how this was written
How we researched this. We read Google's documentation on AI features, generative AI content and spam policies, studies from Pew Research Center, Ahrefs, Seer Interactive, Semrush and SparkToro (the last four sell SEO products, so we flag their findings as vendor data), the Tow Center and EBU/BBC reviews of AI answer accuracy, OpenAI and Pew figures on AI use and Alphabet's results, and compared them with my own experience running a 20-person agency and rebuilding it around AI. We could not open Gartner's own page for its forecast and could not verify later ChatGPT user figures. We will re-check every figure each quarter.
How this was written. Researched with AI tools from the sources below, then checked and edited by Steven Morey.
- Google users are less likely to click on links when an AI summary appears, Pew Research Center (22 July 2025)
- Americans and AI 2026, Pew Research Center (17 June 2026)
- AI Overviews reduce clicks (update), Ahrefs (4 February 2026)
- AI Overviews impact on Google CTR (September 2025 update), Seer Interactive (4 November 2025)
- 2026 zero-click search study, SparkToro (9 June 2026)
- Semrush AI Overviews study, Semrush (refreshed 15 December 2025)
- AI search is driving more queries and higher quality clicks, Google (6 August 2025)
- We compared eight AI search engines. They're all bad at citing news, Tow Center, Columbia Journalism Review (6 March 2025)
- Using generative AI content, Google Search Central (updated 1 October 2026)
- Spam policies for Google web search, Google Search Central (updated 28 August 2026)
- AIs are highly inconsistent when recommending brands, SparkToro (28 January 2026)
- ChatGPT has more than 800M weekly active users, says Altman, Tech.eu (7 October 2025)
- Traditional search forecast to fall 25% by 2026 (Gartner), MediaPost (19 February 2024)
- Alphabet second quarter 2026 results, Alphabet, filed with the US SEC (22 July 2026)
- AI assistants misrepresent news content, says major study (summary of the EBU and BBC research), AIB (23 October 2025)
