How to choose an AI SEO agency in 2026 – what they do, what to demand and what it costs

We prefer to call this way of working AEO – Answer Engine Optimization. Whatever the label, the aim is for your company to be visible, and correctly described, when potential customers put questions to ChatGPT, Claude, Gemini and AI Overviews. Not just ranking in the classic list of blue links – but being cited in the answer itself.

The problem is that the field is young. Many agencies simply put "AI" in front of their regular SEO without changing what they actually deliver. The industry is also divided: one camp argues that AEO is essentially the same as SEO, another that it is a completely new discipline – and how an agency answers that question reveals how well informed it is (our position is further down). This guide helps you decide whether you need an AI SEO agency and how to choose one: how to tell competence from hype, what to demand from the agency and what it costs.

If you want to understand the basics of how AI search works first, we recommend our complete guide to AI SEO. This article focuses on the decision: choosing the right partner.

What does an AI SEO agency do – and how does it differ from a regular SEO agency?

An AI SEO agency optimises for answer engines, not just search engines. The technical foundation is largely the same as in classic SEO – a fast website, correct indexing and quality content – but a few things come on top.

The most important differences from a traditional SEO agency:

• Questions instead of keywords – AI queries are longer and more conversational, and get broken down into sub-questions. An AI SEO agency maps question clusters and prompts, not just individual keywords.
• Entity work – Making sure your company is correctly represented as an entity in the models' knowledge base, so they know what you do and who you are.
• Mentions outside the website – AI answers are influenced by what is said about you elsewhere. Earned mentions become part of the work, not an afterthought.
• Measuring citations – The focus shifts from ranking alone to how often, and how accurately, you are mentioned in AI answers.
• Complete product information – If an AI lacks information about a feature you actually have, it tends to answer "no" rather than "I don't know". That costs you business directly.

Do you need an agency, or is a single consultant enough? It depends on the scope. A consultant may be enough for advice and a clearly defined piece of work, while an agency is usually needed when the work spans technology, content and ongoing measurement at the same time – and then you want an agency that can fill the role of marketing engineer.

Three signs that you actually need help with AI visibility

You do not need to bring in help just because AI search is a hot topic. But three signs suggest it is time.

1. You are not mentioned – or are mentioned incorrectly – when someone asks an AI about your category. Test it yourself: ask ChatGPT and Perplexity twenty or so questions your customers would ask, and record the answers.
2. An AI describes you incorrectly. If information is missing, the model often answers "no" rather than "I don't know" – and the deal is gone before you have even had a chance.
3. Your organic traffic is falling even though you are not losing rankings. The clicks are going to AI answers that answer the question directly. We look at the data behind that shift in How does AI affect SEO?

One sign in the other direction is worth mentioning: if your basic SEO or your website has fundamental flaws, fix those first. AEO built on a broken technical foundation has less to stand on.

Is AI SEO just regular SEO with a new name?

Partly – and that is exactly why you should be sceptical of overconfident answers. There are two camps: one that claims AEO is fundamentally the same as SEO, and one that sees it as a completely new discipline. As far as the data goes, it points both ways at once.

Our position: much of it rests on the same foundation as SEO. Crawlability, technical structure and what we would call classic SEO impact still matter – the models read the web, not a parallel version of it. But three differences are real, and that is where an AI SEO agency adds genuine value:

• The breadth of the questions. Basic keyword research is not enough when questions are asked in free text and broken down into sub-questions. Context and question mapping become the core of the work.
• The quality of the traffic. Someone who clicks through after reasoning with an AI is often further along in their decision – fewer visits, but warmer ones. We look at the data in How does AI affect SEO?
• Not just being visible, but being described correctly. Being mentioned is not enough. The AI has to summarise your product, your values and your offer correctly – otherwise the visibility is ineffective at best and harmful at worst.

The data shows that AI search is, in part, a discovery layer of its own. According to Ahrefs, 28.3% of ChatGPT's most-cited pages have no organic visibility in Google at all – in other words, they are cited even though they do not rank. At the same time, some assumptions have turned out to be weak: when Ahrefs added schema markup to 1,885 pages, AI citations barely moved. The lesson is not that technology is unimportant, but that shortcuts are rarely the real lever. We follow developments and the data closely – as with SEO, the playing field can change quickly. What matters is not which camp an agency belongs to, but that it bases its decisions on data.

The terms around AI search are also used interchangeably. Here are the ones you will come across:

• SEO – optimising to rank in search engines' lists of links.
• AEO (Answer Engine Optimization) – optimising to be cited in AI answers. The term we prefer.
• GEO (Generative Engine Optimization) – often used as a synonym for AEO. We untangle the difference in AI Overviews and GEO.
• AIO – usually refers to Google's AI Overviews, sometimes to AI optimisation in general.
• LMO and LLMO – Large Language Model Optimization. Yet another variant on the same theme: optimising for language models.

Do not be impressed by an agency that leans heavily on acronyms. What matters is what they actually do – not which term they have chosen.

How should a serious agency measure results – and why is rank tracking alone the wrong KPI in 2026?

Ranking is still relevant, but as the only metric it is no longer enough. As more and more questions are answered directly in the AI interface, the value shifts from clicks to visibility in the answer itself. An agency that only reports positions is measuring the wrong layer.

Here is what a serious agency should report instead:

• Mentions in AI answers – how often, and in which questions, your brand is mentioned.
• Share of voice – your share of the AI answers compared with your competitors.
• Cited pages – which of your pages are actually picked up as a source.
• Accuracy – whether the information the model gives about you is correct.
• Business data and attribution – since so much happens without a click, asking "how did you hear about us?" matters more than last-touch attribution.

The weight of mentions is not guesswork. According to Ahrefs' analysis of 75,000 brands, brand mentions correlate 0.664 with visibility in Google's AI Overviews – more strongly than backlinks (0.218) or domain authority (0.326). In a follow-up (May 2026), mentions on YouTube had the strongest correlation of all the factors studied (~0.737), aggregated across ChatGPT, AI Mode and AI Overviews. The conclusion: PR, earned media and brand work are now part of the SEO discipline, not something alongside it.

Tools in use today include Bing Webmaster Tools (a first-hand source of citation data for Microsoft's AI answers), Ahrefs Brand Radar and systematic prompt testing. Insist that the agency shows how it measures – not just that it does.

The questions that reveal hype or competence

A short conversation is usually enough to tell whether an agency knows its AEO. Ask these questions – and listen just as carefully to what they do not promise:

• Can you show a concrete example of getting a client's brand cited in AI answers?
• How do you measure results, and which KPIs do you report every month?
• How does your methodology differ from regular SEO with "AI" on top?
• What do you promise – and what do you deliberately avoid promising?
• How do you work on mentions outside our own website?

Some answers are clear warning signs:

• Guarantees of placement in AI answers. AI generates answers based on context and model behaviour that no agency controls. Guarantees are the biggest red flag.
• Rank tracking as the only KPI. It shows the agency has not adapted to how AI search actually works.
• llms.txt or FAQ schema as a miracle cure. Several of the tactics sold as decisive are still unproven. Be sceptical of anyone who promises a big effect from a single technical file.
• Best-of lists where the agency itself comes out on top. A "best AI SEO agencies" list that happens to be published by one of the agencies is marketing, not an independent assessment.

What should you demand from the agency in the first 90 days?

Serious work starts by measuring where you stand, not by producing content. This is what you can reasonably expect in the first three months:

1. AI visibility audit – a picture of how and where you are mentioned today in ChatGPT, Perplexity, Gemini and AI Overviews, and which competitors own your questions.
2. Entity and question mapping – a prioritised list of the questions your customers ask AI and the answers you need to own.
3. Structure and content – pages structured so they can be cited: clear answers, semantic chunking and correct structured data.
4. A plan for earned mentions – how you will become visible in the sources the models draw on.
5. A monthly report with the right KPIs – mentions, share of voice and cited pages, not just positions.

Demand a clear method and realistic timelines. The first measurable effects usually appear within a few months – not weeks – and the work is ongoing, not a one-off project.

What does an AI SEO agency cost?

There is no standard price, but the structure usually follows the same pattern: an initial strategy and optimisation phase followed by ongoing work on a retainer. The price depends on how mature you are, the competition in your industry and the scope of the work. We set out actual price levels in Sweden and the three most common pricing models – hourly rate, retainer and fixed project – in What does an SEO agency cost in 2026? An AEO engagement is structured the same way, but the emphasis shifts towards strategy, entity work and measuring AI visibility rather than pure link and content production.

Be wary of the extremes. A price that seems too low usually means standardised tasks with no strategy behind them. A very high price means the agency should be able to show exactly what is included and how it is measured. Always ask for a quote tied to clear deliverables.

How to take the next step

Choosing the right partner is less about which terms the agency uses and more about what it delivers and how it measures it. Want to see where your company is visible – and not visible – in AI answers today? Get in touch and we will analyse where you stand. You can also read more about how we work as an AI SEO agency.