Get found where people actually look.

AI assistants add a fourth behaviour. They retrieve, summarise and attribute. When someone asks an assistant which dentist in Oak Brook does same day crowns, the assistant assembles an answer from sources it can read and trust, and names a few of them. If your site is thin, contradictory, or unreadable without JavaScript, you will not be one of the names.

None of this is mysterious, and very little of it is secret. Google publishes what it wants, what it discourages and what it forbids. The difficult part is not knowing the rules. It is doing several hundred small, unglamorous things consistently for long enough that the compound effect shows up in the numbers.

AEO, answer engine optimization, is the narrower craft of structuring a page so that a clean, correct answer can be lifted out of it: a direct question as a heading, a two or three sentence answer immediately underneath, no throat clearing, and structured data that matches what is visible on the page.

What these three share is that they reward the same underlying thing: a site that is easy to read, factually consistent with itself and with your listings, and genuinely useful about a narrow subject. What differs is the shape of the page and how you know it worked.

One honest consequence. If an assistant answers the question fully, the person may never click. That is a real cost of this shift, and any agency that pretends AI visibility is pure upside is not being straight with you. What it buys you is being the named business in the answer, which for a local service business is often worth more than the click was.

The numbers that mean something are further down the chain. Calls that lasted long enough to be a real conversation. Forms that produced a booked appointment. Direction requests and profile calls from the Business Profile. Branded search volume over time, which is the cleanest proxy for whether more people know you exist. Revenue by channel, if your systems can carry it that far.

We report against a small set of those and against the specific things we said we would do that month. If the work did not move the number it was supposed to move, the report says so, in the first paragraph.

What counts as a realistic target depends heavily on the market. Density changes everything: a dental practice in a Chicago suburb competes inside a radius of a few miles against practices with the same services and similar reviews, so proximity and profile quality dominate. The same practice type in a rural county may rank in the map pack almost by default, and the effort belongs in content and conversion instead.

Regulated categories change the work rather than the measurement. Healthcare copy has to avoid outcome claims, which removes the easiest persuasion tools and forces genuinely informative content. Legal marketing sits under state bar advertising rules that differ by state. Financial services carry their own disclosure requirements.

Purchase cycle changes what content is for. Emergency categories such as water damage restoration or urgent care are dominated by intent that appears and resolves within an hour, which makes speed, proximity and phone answering worth more than a blog. Considered purchases such as orthodontics, roofing or a legal retainer involve weeks of research, which is where depth of content actually pays.

Seasonality is real and usually ignored. An HVAC company that starts search work in June sees the first honest read the following spring. Building in the shoulder season is the point, not a delay.

Reputation and CRM sits on the other end. Reviews feed map pack visibility directly, and follow up decides whether the enquiries that search produces turn into booked work. A practice with strong rankings and no follow up loses most of what it earned.

Branding and social media matter here in a less obvious way: branded search volume is one of the cleanest signals that more people know who you are, and it is fed by everything that is not search.