Is AEO worth doing in 2026?

The underlying work is worth doing, because being visible in AI answers is a real and growing surface. But most of what makes you visible in AI answers is good SEO and good PR: useful, specific content on a crawlable site, plus credible third-party coverage. So yes, care about it, but treat it as part of SEO and digital authority rather than a separate channel with its own budget. The thing not worth doing is paying a premium for AEO as a distinct service when the deliverables are the SEO and content work you should already be doing.

Can you actually measure AI search visibility?

Not in the way you can measure classic search, and that is the central problem. Classic search has a stable, countable unit, the query, and a real denominator in impressions, so click-through rates mean something. Conversational AI queries do not repeat, do not match in tidy phrases, and have no observable denominator, and the engines are non-deterministic, so the same question can return different answers and sources. Tracking tools give you a sample of citations over a prompt set someone chose, which is a useful diagnostic but not a clean measurement of a channel.

What do tools like Profound and Scrunch actually do?

They monitor multiple answer engines by running prompts and capturing the responses, then report how often your brand is mentioned or cited, how that compares to competitors as a share of voice, the sentiment of how you are described, and which sources the engines rely on. Some add auditing and content suggestions. That is genuinely useful as a diagnostic. The limits are that the data is a sample over a chosen prompt set rather than a population, the engines barely agree on who to cite, and you usually cannot trace a citation through to a business outcome.

Why is it hard to measure AI citations without bias?

Because the measurement starts with a prompt set that someone designs, and that design encodes assumptions. If you build the prompts from what you think people would search for, you import a keyword mindset into a medium that does not behave like keywords, so the data reflects your assumptions about how people ask rather than how they actually use LLMs conversationally. Change the prompt set and the numbers change. When the result depends that much on the design of the test, you have a reflection of your own assumptions more than an objective metric.

Is AEO just SEO with a new name?

Largely, yes, for the on-site work, and Google now says so in its own guidance: optimising for its generative AI features is still SEO, because those features run on the same index and ranking systems as classic search. The genuine addition is an off-site, brand-perception layer, how credibly and consistently third parties describe you, which feeds the sources engines draw on. But that layer is PR and digital authority, not a new optimisation science. So treat AEO as the SEO and PR you should be doing, pointed at AI surfaces, rather than a separate discipline to buy.