AI SEO Services: What You Are Actually Buying

Two agencies can sell you AI SEO services this week and deliver work with almost nothing in common. One will point a language model at your blog and hand you forty articles by Friday. The other will spend that same week on your structured data, your crawler access and how clearly your site explains what your business actually is. Same phrase on the invoice. Opposite products.

I am Armen, an SEO consultant in Barcelona, and this is the confusion I unpick most often on a first call. Someone has been paying for AI SEO for six months, the monthly report is full of activity, and their business still does not exist when a customer opens ChatGPT and asks who to hire. They bought the automation and assumed they were buying the visibility.

So here is the buyer’s version, including the part that argues against signing anything.

What are AI SEO services?

AI SEO services are paid work aimed at your visibility inside AI driven search: the AI Overview sitting above Google’s results, and the answers ChatGPT, Gemini, Perplexity or Copilot give when someone asks them for a recommendation. The goal shifts from a ranked blue link to being the business the answer names.

The reason anyone is buying this is sitting in the click data. In the first four months of 2026, 68.01% of Google searches ended without a click at all, according to SparkToro’s analysis of Similarweb clickstream data, up from 60.45% two years earlier. Ahrefs found the top ranking page loses roughly 58% of its click through rate when an AI Overview appears above it. Pew Research measured the same effect from the user side: with an AI summary present, people click a result on about 8% of visits instead of 15%.

Your ranking did not move. Your traffic did.

Why does AI SEO mean two opposite things?

Because the phrase was claimed by two industries at once. One group sells AI as the machine that produces your SEO. The other sells AI as the place your customers now search. Both write “AI SEO services” on the landing page, and the buyer is left to work out which one arrived.

Neither is a scam. Automation genuinely lowers the cost of a page, and any decent team uses AI somewhere in the process now. The problem is only the label, and the label is doing a lot of quiet work on those sales pages.

One caution on the production side. Volume is the one thing language models are least impressed by, because forty articles assembled from the same public sources give a model nothing it cannot already write itself.

Which one does your business actually need?

Most businesses I speak with need the visibility work, not the production work. Production help earns its keep only when publishing is genuinely your bottleneck, which for a local service business in Barcelona it almost never is. Your bottleneck is that four competitors are already in the answer and you are not.

Look at where your customers actually are before you buy either. AI referral volume is still small next to Google, by a wide margin, and anyone telling you Google is finished is selling something. What has changed is the quality of that small stream. Analyses through 2026 keep finding AI referrals converting several times better than non branded organic, with ChatGPT accounting for the large majority of it, because the visitor arrives already briefed and mostly decided.

That is the honest case for funding this now. Not the volume. The fact that a handful of visitors a week arrive halfway to buying, and the fact that being cited gets harder once your competitors are established in the answer.

What does a real AI search engagement include?

It starts with access and ends with mentions. If a model cannot fetch your pages, everything else is decoration, so the first job is your robots.txt and your deliberate decision about which AI crawlers you allow. Plenty of sites block the bots by accident and then wonder why they never appear.

After that the work is unglamorous:

1. An entity layer that states plainly who you are, what you sell and where you operate, consistent across your site, your Google Business Profile and the directories that describe you.
2. Structured data on the pages that matter. Google is clear that schema is not required for generative AI features, but it remains the cheapest way to make a page unambiguous, and it earns you rich results either way.
3. Content restructured so a clean answer can be lifted out of it. A model quotes a passage, not a page, which means the answer to the question in your heading belongs in the first two sentences under it.
4. Bing indexing, since a large share of what ChatGPT and Copilot fetch live runs through that index and most Spanish businesses have never opened Bing Webmaster Tools.
5. Presence on the sources models read when they need a recommendation: review platforms, sector directories, local press, and the Reddit threads that quietly decide a lot of these answers.

Google’s own optimisation guide says the quiet part out loud: optimising for generative AI search is optimising for the search experience, and thus still SEO. That single sentence should reset the price of anything sold to you as a brand new discipline. It is the same SEO consulting work, pointed at a different surface.

How do you measure AI visibility without inventing numbers?

Three sources, and none of them costs anything. Search Console now carries a generative AI performance report showing impressions and clicks earned inside AI experiences. Your analytics will show referrals from chatgpt.com and perplexity.ai as their own channel. And a fixed list of prompts, re run monthly, tells you what a customer actually sees.

That last one is the most useful and the most neglected. Write down the eight questions a real buyer would type, ask them across two models on the first Monday of every month, and record who gets named. It takes twenty minutes and it is more honest than most dashboards.

Be wary of the scores. Google has explicitly warned that no third party tool has access to its internal AI metrics, so when a report hands you an AI visibility index of 74 out of 100, the only question worth asking is what it counted. If nobody can answer that in one sentence, it is a number designed to be renewed rather than understood.

What do AI SEO services cost?

Published ranges tell you more about positioning than about value. United States agencies commonly quote from around 2,500 dollars a month for a small business up past 15,000 for enterprise work. Retainers here in Spain sit well below that, and a full SEO audit is often sold as a fixed price piece instead.

The better question is whether it should be a retainer at all. A meaningful chunk of AI visibility work is a defined set of fixes with an end: the crawler access, the entity data, the schema, the restructured answers. That is a project. What genuinely continues afterwards is content, authority and the monthly check on what the models say, and that is ordinary SEO work with a new report attached to it.

I would rather sell someone a scoped piece of work that finishes than a permanent AI retainer for a problem that took a month to fix.

How can you tell which kind of vendor you are talking to?

Ask four questions and the answer arrives fast.

Which AI crawlers can currently reach my site, and how did you check? A visibility vendor has looked at your robots.txt before the call. A production vendor will talk about content volume instead.

Where will you measure this? If the generative AI report in Search Console and your own referral data do not come up, the measurement is going to be their own invented score.

Who writes the content, and what do they know that a model does not? Bulk generated pages are the fastest way to publish and one of the slowest ways to be cited.

And the one that sorts most of it: what happens when this stops working, because the models change every few months? A serious answer describes a monitoring routine. A weak one promises a permanent method.

What I would buy first

If I were spending my own money on this, I would not start by hiring an AI SEO service at all. I would spend one afternoon checking that the AI crawlers can reach my site, that my business is described identically everywhere it appears, and that my most important pages answer their own headings in the first two lines. Most of the businesses I audit gain more from that afternoon than from their first two months of any retainer.

Then I would buy the diagnosis before the subscription. A scoped audit tells you which of the two products you actually need, and it means you sign the monthly agreement knowing what it is for, if you sign one at all.

I am Armen, an SEO consultant in Barcelona, and I run ACERO Digital with a team who spend a good part of every week on exactly this problem. If you want to know where your business currently stands in AI answers before anyone sells you a package, tell us about your project and we will take a look at what the models say about you today.