Reference · updated August 2026
AI answers your customers now. It rarely mentions you.
Search used to give ten links. AI gives one answer and names a few sources. This site explains, in plain terms, how those sources get picked. Every number here comes from an outside study and links to it.
What changed
People stopped clicking
When AI shows a summary, far fewer people click anything at all.
Read those last two together. Being cited is good for influence, but it will not send you much traffic. And the clicks you lost are not showing up somewhere else. They are simply gone.
Users clicked a traditional result on 8% of searches where an AI summary appeared, versus 15% where none did: a drop of roughly 47%.
Study of 68,000 real search queries tracking actual user behaviour.
Here is the awkward part. If AI never mentions you, nothing in your analytics will tell you. No lost ranking, no drop in impressions. You just quietly stop being considered.
The basics
Three terms, three different jobs
Vendors use these words loosely. Here is the difference that matters.
| What you get | How you measure it | |
|---|---|---|
| SEO | A link in a list, and a click | Rankings and visits |
| AEO | A citation inside the AI answer | How often you get cited |
| GEO | An accurate description of you | How often you are mentioned, and whether it is right |
In short: SEO gets you a click. AEO gets you quoted. GEO makes sure what it says about you is correct.
Two clocks, not one
AI answers in two ways. Sometimes it searches the web and answers from what it finds. That responds to what you publish, within days or weeks.
Other times it answers from memory, using its training data. That only changes when the model is retrained, which takes months. Publishing more on your site will not move it.
Most programmes get cancelled because someone measured the fast thing against the slow clock, and it looked like nothing was working.
The research
What actually works
One proper study underpins most of this field. Researchers tested content changes across about 10,000 questions and measured which ones got cited more.
Targeted content changes raised visibility in generative engine responses by up to 40%.
GEO-bench: roughly 10,000 queries across nine datasets. Authority-style edits, adding citations, statistics and quotations, outperformed cosmetic edits such as rewriting, simplification and keyword work. Effect size varied by domain.
The finding is simple. Evidence beats polish.
Adding real numbers
Say '34 markets since 2019', not 'extensive experience'. This was the strongest single change they measured.
Quoting named sources
AI can safely repeat a sourced claim. It has no reason to repeat an unsourced one.
Stuffing keywords
This made things worse, not better. The old SEO reflex works against you here.
Where citations go
AI mostly cites other people
This is the uncomfortable one for anyone planning to win citations with their own website.
Roughly 85% of AI references point to third-party platforms rather than brand-owned sites.
Most citations point at big community and reference sites, not company websites. A small group of domains takes most of them.
The top 15 domains account for roughly 68% of all citations produced by the five major AI answer engines.
So for most companies the realistic goal is not to become a cited site. It is to be described accurately on the sites that already get cited.
Watch out
Three things being sold that do not work
llms.txt as a ranking factor
Google says it has no effect. No major AI company reads it. One study of 137,000 sites found 97% of these files got zero traffic. Harmless to publish, but it is not a lever.
Guaranteed placement in AI answers
Nobody controls this. Anyone promising a spot is selling something they do not own.
A single 'AI visibility score'
It blends the two clocks above into one number. You cannot act on it.
Google documentation updated in June 2026 states llms.txt has no effect on Search rankings or AI Overviews. John Mueller noted that no AI crawler has claimed it extracts information from the file.
Start here
Four steps, in this order
-
01
Go and look
Ask the five questions your buyers ask, in the AI tools they use. Write down who gets named. Most companies have never done this.
-
02
Make sure AI knows who you are
If it cannot tell your brand name, legal name and website are one company, nothing you publish counts for you.
-
03
Fix your best pages first
Put the question in the heading. Answer it in the first two lines. Add real numbers. Cheaper than writing new pages, and better supported by the research.
-
04
Go where the citations already are
Since most citations point elsewhere, being described correctly on those sites beats publishing more on your own.
Sources
Where these numbers come from
Every figure on this page is external and linked. The studies measure different populations over different windows, so read them as converging evidence rather than one authoritative number.
- Pew Research CenterUsers clicked a traditional result on 8% of searches where an AI summary appeared, versus 15% where none did: a drop of roughly 47%. searchengineland.com ↗
- Pew Research CenterWhen an AI Overview is shown, users click one of its cited source links about 1% of the time. searchengineland.com ↗
- Search Engine LandGoogle zero-click searches reached 68% in early 2026. searchengineland.com ↗
- Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan & Deshpande, ACM SIGKDD 2024Targeted content changes raised visibility in generative engine responses by up to 40%. arxiv.org ↗
- Omnibound, AEO statistics compilation 2026Roughly 85% of AI references point to third-party platforms rather than brand-owned sites. www.omnibound.ai ↗
- AI Platform Citation Source Index 2026The top 15 domains account for roughly 68% of all citations produced by the five major AI answer engines. everything-pr.com ↗
- Google Search RelationsGoogle documentation updated in June 2026 states llms.txt has no effect on Search rankings or AI Overviews. John Mueller noted that no AI crawler has claimed it extracts information from the file. baselinelabs.ai ↗
Keep reading
FAQ
Common questions
What is Answer Engine Optimization?
It is the work of making your content the thing an AI assistant quotes when it answers a question, instead of competing for a spot in a list of links.
Is SEO dead?
No. Google still sends about 87.6% of website traffic. All AI assistants combined send about 0.29%. What changed is that many searches now end without a click, and you cannot see that in your analytics.
Is AEO the same as GEO?
They overlap but aim at different things. AEO earns the citation. GEO controls how the AI describes you, whether or not it links to you. A citation attached to a wrong description is worse than none, because it now carries the AI's authority.
How long does this take?
If the AI searches the web, new content can show up within days or weeks. If it answers from memory, you are waiting for the next model to be trained, which takes months. Anyone quoting one timeline for both has not separated them.
Do I need an llms.txt file?
There is no evidence it helps. Google says it has no effect on Search or AI Overviews, and no major AI provider reads it. It is cheap and harmless, but it is housekeeping, not strategy.
Who runs this site?
Lifewood Data Technology, an AI data company that also sells AEO and GEO services. Every number here links to an outside source so you can check it without trusting us.
Want to show up in AI answers? <span class='tone'>Lifewood can help.</span>
We measure where your brand stands in AI answers today, then fix what is blocking you. Start with a baseline audit before spending anything on content.
- 100+ languages
- 50+ delivery centres
- Measured monthly
Lifewood Data Technology publishes this reference.