The honest version of this story is duller than either side of the argument wants it to be. Fewer people will click through to your website from search, and a higher proportion of the ones who do will already be most of the way to a decision. Those two things arrive together. I run Growth Bully, a Malta performance marketing agency, and the most common mistake I see is a business reading the first half of that sentence, concluding it is failing, and changing course while it is actually improving.
This is the educational half of the story. If you want the mechanism, meaning how these systems decide which businesses to name in the first place, my team has written that up at thegrowthbully.com/bytes/how-ai-assistants-pick-who-to-recommend. What follows is about what changes for the business, not how to game it.
Does AI search reduce my website traffic?
Usually yes, and most of the loss is traffic that was never worth much. Questions with a single factual answer now get answered in the results. Your opening hours, what a term means, whether you serve a particular area: all of that used to arrive as a session and leave immediately. It still happens, just without the visit.
What that does to a traffic chart is ugly and misleading. Total sessions fall, average time on page often rises, and the pages that lose the most are the ones that were converting least. If sessions are the number your month is judged on, this reads as a crisis. If enquiries are, it frequently reads as nothing at all.
What is zero click search, in plain terms?
It is a search that gets answered without anyone visiting a website. The answer is assembled from sources, shown in place, and the person moves on satisfied. It is not new, but AI answers have widened it from simple facts to questions that used to need a page: comparisons, how something works, what to watch out for.
The consequence for a small business is specific. You can be the source behind an answer that produces no visit at all, which means you were useful, were cited, and saw nothing in your analytics for it. That gap between influence and measurement is the genuinely new problem, and nobody has solved it cleanly yet.
What stops working?
Thin informational content written to capture clicks. Short pieces that define a term, list five tips, or answer a question that has one correct answer. Those pages existed to intercept a search and funnel the visitor onward, and that interception is exactly what the answer layer now does better.
The categories I would stop investing in:
- Definition pages that explain a term with no experience attached. The answer engine can do this, from your page, without sending anyone to it.
- Thin listicles assembled from what already ranks. Nothing in them is yours, so nothing in them is worth citing.
- Keyword variants of the same page. Three near-identical pieces used to catch three phrasings. One substantial piece now covers all three, and the duplicates dilute it.
- Anything written for a robot’s benefit. If a human would not finish it, it will not be chosen as the basis for an answer either.
None of that means publishing less. It means publishing fewer things that only ever worked as a toll booth.
What starts mattering more?
Being described accurately in places you do not control. These systems build a picture of your business from your own site plus directories, listings, reviews, press, social profiles and forum threads. When those sources disagree about what you do, where you operate or whether you are any good, the answer gets hedged or you get left out.
That is an uncomfortable shift because it is mostly not on your website. A stale directory entry describing a service you dropped two years ago is now a live liability rather than a tidy-up job. So is a profile with your old address, or a category that says you do something you do not. I have seen more damage from inconsistency than from absence.
The second thing that matters more is first-hand experience. Specific numbers from work you actually did, methods you can explain, and opinions that could only come from having done the job. Not because it pleases an algorithm, but because it is the only kind of content that cannot be assembled from what already exists.
If you want the practical optimisation side of this, that is covered in answer engine optimisation, and the paid search implications sit in AI Overviews and Google Ads.
Which number should I be watching instead?
Enquiries, and the quality of them. Count the conversations that start and what share of them are the right kind of business, then compare that to the same period last year. Sessions become a diagnostic rather than a goal: useful for working out why enquiries moved, useless as the thing you optimise.
In practice that means two changes. Someone has to record where enquiries claim to have come from, because the analytics will increasingly not tell you. And the monthly conversation has to stop opening with traffic, which is harder than it sounds when the traffic chart is the one everybody already knows how to read. There is a sceptical guide to reading those reports at thegrowthbully.com/bytes/how-to-read-your-marketing-report.
How quickly does any of this happen?
Slowly, and unevenly by industry. Informational queries have already moved. Decisions that involve spending real money with a local supplier still mostly end in a human visiting a website, reading about the business, and making contact. The direction is clear and the timescale is years, not months.
I would be sceptical of anyone, including me, who puts a date on it. What I would not be sceptical of is the direction, because every platform involved has the same commercial incentive to answer the question rather than hand the visit away.
What would I actually do about it?
Three things, in order, and none urgent in the way the headlines suggest. Fix what the internet says about your business, because it is cheap and wrong more often than owners expect. Shift your content towards what could only come from you. Change the number your month is judged on.
That last one is the hardest, and it is the one that decides whether the next two years read as decline or as improvement. Change it deliberately, before the traffic chart makes the decision for you.
What I would not do is panic-publish, buy a tool that promises AI search visibility, or rewrite a working website around a transition that is still in progress. The businesses that come out of this well will mostly be the ones that were already accurate about themselves and specific about their work, which is a very old answer to a very new question.
The commercial version of this argument, including how the agency approaches it, lives at thegrowthbully.com/bytes/aeo-for-malta-businesses.
Frequently Asked Questions
Will AI search kill my website traffic?
No, but it will change its composition. Expect fewer visits overall, concentrated on questions that genuinely need a page, with the simple factual queries answered before anyone clicks. Watch enquiries rather than sessions for a few months before concluding anything, because the two numbers now move in different directions.
Should a small business still write blog posts?
Yes, but fewer and better ones. Stop writing pieces that only define a term or repeat what already ranks, because the answer layer does that without sending a visit. Keep writing what comes from having done the work: specific results, methods, judgement calls, and the mistakes you would not repeat.
How do I find out what AI assistants say about my business?
Ask them, in the wording a customer would use, and read the answer critically. Note whether you are named at all, whether the description matches what you actually sell, and which sources the answer leans on. Repeat it monthly. It is crude, free, and more informative than most paid tools.
Is it worth paying for AI search optimisation tools?
Not yet, for most small businesses. The highest return work is unglamorous and manual: correcting your listings, making your own pages accurate and specific, and collecting real reviews. Tools that promise visibility in AI answers are mostly reporting on a surface nobody can reliably influence directly.
Does this mean SEO is finished?
No, it means the payoff changed. Being crawlable, fast, accurately described and genuinely useful still decides whether you can be chosen as a source. What has weakened is the specific tactic of ranking a thin page to capture a click, which was always the least durable part of the discipline.



