If a growing share of searches end without a click, the honest question is what your SEO budget is still buying. The answer divides sharply by content type. Simple informational content is being absorbed by AI answers and will not come back. Content requiring judgement, comparison or trust is largely unaffected, because the AI answer sends people to it rather than replacing it.
What is actually at risk
Not SEO as a whole. A specific and identifiable category of page.
| Content type | Exposure to AI answers | Why |
|---|---|---|
| Definitions and simple facts | Very high | One paragraph fully satisfies the query |
| Basic how-to steps | High | The answer engine can list the steps itself |
| General category overviews | High | Nothing specific enough to require the source |
| Comparisons with real trade-offs | Moderate | Buyers want to verify and see the reasoning |
| Supplier and vendor evaluation | Low | Choosing someone requires assessing them directly |
| Local specifics and pricing context | Low | Varies enough that a general answer is unreliable |
| Original data and observation | Lowest | The answer engine needs a source, and this is it |
The pattern is consistent: the more a page contains that could only come from doing the actual work, the less an AI summary can replace it — and the more likely it is to be cited as the source.
The reallocation, not the reduction
The instinct when a channel looks threatened is to spend less on it. Here that would be the wrong response, because the underlying work has not lost value — the output it should aim at has changed.
- Move spend away from volume. Twenty thin articles a month aimed at informational terms was always weak and is now actively pointless.
- Move it towards specificity. Fewer pages, each containing something an answer engine would need to quote: real numbers, considered comparison, genuine observation.
- Keep funding the technical foundation. If AI crawlers cannot render your pages, none of this matters. This has become more important, not less.
- Fund authority. Answer engines weight source credibility heavily. Named authors with real credentials and earned mentions matter more now than they did.
What we changed about our own publishing
We publish less than we used to and spend longer on each piece. The article you are reading exists because we went and measured something in our own category rather than restating what everyone already says about AI search.
That is not a virtue claim, it is the mechanism: original observation is the one thing an answer engine cannot generate for itself, so it is the one thing reliably worth publishing. Everything else we might have written about AI search already exists in a hundred versions, and an answer engine has no reason to prefer ours.
The uncomfortable arithmetic
For some businesses the honest conclusion is that their content programme was never producing much, and AI answers have simply made that visible. Traffic to informational pages that never generated an enquiry was always a vanity metric; losing it costs nothing real and reveals what the programme was actually delivering.
That is worth checking before deciding AI search has damaged you. Look at whether the traffic you are losing was ever converting. Often the answer reframes the whole problem — and occasionally it means the budget was misallocated for years and this is the prompt to fix it.
If the traffic you are losing to AI answers was never converting, you have not lost revenue. You have lost a number that was flattering you.
Where this settles
Search demand has not fallen. People still need suppliers, still compare options, still want to know what something costs and whether it will work for them. What has changed is which questions get answered without a visit.
So the budget still has a job. It is buying a smaller number of genuinely useful, specific, credible pages instead of a large number of forgettable ones — which is the version of the work that was always going to last.
Frequently asked questions
Will AI search reduce my website traffic?
For informational queries, very likely — those are the searches AI answers satisfy completely. Commercial queries behave differently, because someone choosing a supplier still needs to evaluate options, and that means visiting sites.
Should I reduce my SEO budget because of AI search?
Reallocate rather than reduce, in most cases. The work that earns AI citations is largely the work that earns rankings. What deserves cutting is thin content aimed at informational terms that AI now answers outright.
Which content is most at risk from AI answers?
Simple factual and definitional content — the kind of page that answers a question in two sentences. If a single paragraph fully satisfies the query, the AI answer will provide it and the visit will not happen.
Which content is safest?
Anything requiring judgement, comparison, local specificity or trust. Choosing a supplier, understanding a trade-off, or evaluating options against your own circumstances still sends people to actual pages.
How should I change what I publish?
Fewer pages, more specific. Publish things an AI answer would need to cite rather than things it can replace: original observation, real numbers, considered comparison, and material only someone doing the actual work would know.
If you want to know which of your pages are exposed and which are safe, send us the domain. We will tell you where we think the risk actually sits. See where we think budget belongs, message us on WhatsApp, or send an enquiry.
Sources
Written by Charlotte Zhang, Operations Partner at Kaizenaire Pte Ltd (UEN 201932071D). Charlotte co-founded Kaizenaire with Ken Tan in 2019 and runs client delivery, after four years in operations at OCBC and HSBC.