A growing share of your future customers will never see a search results page on the way to your competitor. They’ll ask an assistant — ChatGPT, Gemini, Claude, the voice thing in their car — “who should I use for X near me?” and receive a shortlist of two or three names.

The discipline of getting onto that shortlist has picked up a name: answer engine optimization (AEO — you’ll also see GEO, for generative engine optimization; the terminology war is ongoing and unimportant). The work behind the name is concrete, and most of it is doable this quarter.

How machines choose who to recommend

Nobody outside the AI labs knows the exact recipe, but the observable pattern is consistent and reassuringly boring. Assistants recommend businesses that are:

  1. Legible — the AI can determine what you do, where you do it, and for whom, without inference gymnastics.
  2. Corroborated — what your site claims matches what Google Business Profile, directories, and reviews say about you.
  3. Reputed — review volume and sentiment weigh heavily, especially for local queries.
  4. Citable — pages that answer questions directly, structured so a machine can quote them cleanly.

Notice what’s absent: tricks. There is no keyword-stuffing equivalent for AEO yet, and given that these systems read language better than they read markup games, there may never be. The lever is clarity.

The work, in priority order

Say what you do, in plain sentences, on pages machines can parse. Your homepage should state your services and service area in actual prose — not exclusively in a hero image or a slogan. If a competent stranger couldn’t summarize your business from your homepage text alone, neither can a machine. Every service deserves its own page that answers cost, process, and scope questions directly.

Ship structured data. Schema.org markup — Organization, Service, FAQPage, LocalBusiness — is the difference between an AI parsing your prose and an AI reading your facts off a form you filled out for it. This is the single highest-leverage technical task, and most small-business sites still have none. (The sites we build for clients ship it by default.)

Adopt llms.txt. An emerging convention: a plain-text file at your site root that tells AI systems what your business is and where the important pages live. It costs an hour and signals machine-readability across the board.

Get your review engine running. For “near me” recommendations, assistants lean hard on review data. A steady drip of genuine reviews (ask every happy customer, make it one tap) beats a burst of ten from your cousins.

Reconcile your citations. Same business name, address, phone, and description everywhere — site, Google Business Profile, directories, socials. Contradictions make machines hedge, and hedging machines omit you.

What about traditional SEO?

It’s not dead; it’s the substrate. Answer engines are trained on and retrieve from the same web Google crawls, and a site that ranks tends to be a site that gets cited. Fast pages, clean structure, genuine expertise — the fundamentals now pay out through two channels instead of one.

The window here is the fun part: your competitors are almost certainly not doing any of this yet. Early clarity compounds — the sources machines learn to cite keep getting cited.

Want to know how legible your business is to the machines your customers are starting to trust? We’ll tell you straight.