AEO Is Not SEO. Here Is What Actually Changed.

Most of what I read about answer engine optimization treats it as SEO with a new coat of paint. Write better content, add some schema, wait for the AI to notice you. That framing had us watching the wrong dashboard for longer than I would like to admit.

I run marketing at CodaPet, in-home end-of-life care for pets, across 170+ markets. Short, local, urgent searches in a category nobody browses for fun. Roughly 80% of our volume arrives through channels we own rather than paid, so when Google changed what a search result physically looks like, it landed on the part of the business I care most about.

Here is what actually happened.

The free inventory did not get worse. It got smaller.

Our Google Business Profile rankings improved. Our impressions fell anyway.

Those two numbers are supposed to move together. They stopped moving together when AI Overviews took over the top of the page. The Map Pack still exists. It shows up less often, further down, on fewer queries. We were getting better at competing for inventory that was quietly being retired.

The standard SEO instinct there is to go fix the ranking. There was nothing to fix. The ranking was fine. The surface shrank, and no amount of local work brings a shrinking surface back. Once I accepted that, the question stopped being "how do we restore this" and became "where does that demand go now, and are we sitting there when it lands."

That is the first thing AEO is not. It is not a recovery plan for organic. It is an allocation decision.

Being ranked and being named are two different wins

We checked how AI answers were describing our category across our markets. In one of them the answer named an individual practitioner rather than the company, surfaced a competitor we had not seen in that position before, and gave us a couple of passing mentions.

We ranked. We were not the answer.

That gap is the whole discipline. Classic SEO competes for a position in a list, and every entry in the list gets a link. A synthesized answer does not behave like a list. It picks. It names one or two entities, describes them in its own words, and everything else becomes context the reader never scrolls to.

So the unit of victory moved. It used to be a ranked URL. Now it is whether a model can confidently say who you are, and trusts that description enough to put your name in the sentence. That is entity work, and most of it is not on your page. It is the same facts stated consistently everywhere you appear, third parties corroborating them, structured data wiring your identity together instead of leaving a model to infer it from fragments.

You cannot bid on that and you cannot ship it in a sprint. It is also the openest lane I have seen in a while, because almost nobody is working it on purpose.

If your own organic picture has gone strange and the usual explanations are not holding, this is the sort of thing I unpack on a [20-minute call](/contact).

The metric that lies in this era

The thing that nearly sent us in the wrong direction was a rate.

Organic sessions grew much faster than bookings. Conversion rate dropped hard. On a dashboard that reads as a crisis, and the tempting story is that AI Overviews intercepted our buyers and left us the tire kickers.

Absolute bookings were up.

What had happened is that our content started ranking for a wide band of informational questions it never used to rank for. Real people, real reads, zero ad cost, and much lower odds of booking that week. Blended in with our city pages, they flattened the rate. The rate had stopped measuring performance. It was measuring traffic mix.

I do not think a blended organic conversion rate survives as a headline KPI right now. Not because the math is wrong, but because AI search is changing the composition of what arrives faster than the rate can mean anything. We look at absolute bookings, and we look at the transactional pages separately from the informational ones.

What we actually changed

We stopped defending a capped channel and moved the work to the pages catching the displaced traffic. Somebody who no longer sees a card with a phone number and a booking button lands on a city page instead. That page now has to do the job the card used to do, on a phone, before any scrolling happens.

We treated the informational flood as an audience instead of a rate problem, and built real paths from it into the pages that convert.

We leaned harder on paid for the transactional queries while that shift was underway, which is exactly why I do not run these as competing engines. The owned side kept blended acquisition cost low enough that we could push paid into the gap without flinching, and the paid data kept telling us which language people actually use when they are ready.

And we worked on being describable. Same facts everywhere, structured, corroborated by sources we do not control.

None of that is a growth hack, and none of it produced a clean before-and-after chart I can wave around. It is the difference between optimizing to get a link and optimizing to be the answer, and I was slow to stop doing the first one.

If your organic numbers look wrong this year, check what changed in the mix before you change the tactics. Usually the thing that broke is a rate, and the thing that actually changed is what a search result looks like.