The alphabet · GEO

Get named in the answer.

The layer nobody else shows you: whether AI can read your site at all — and whether it says your name when your customers ask who to call.

GEO pillar four of four · 25% of your score

Generative engine optimization — being named in AI answers. It starts with access (can GPTBot and ClaudeBot even fetch your pages?), runs through entity clarity (does AI know exactly who you are?), and ends in citations: how often ChatGPT, Perplexity, and Gemini recommend you when your market asks.

See also: Local SEO · SEO · AEO · AIO

Why it matters

Three truths about GEO.

01

The answer is the new page one

Assistants return one or two names, not ten links. Being in that shortlist is a winner-take-most game — and today almost nobody knows whether they’re in it.

02

Most blocks are accidents

Firewalls, bot protection, and CDN rules silently drop AI crawlers every day. The business owner never sees it — AI simply has nothing to read, and recommends someone it can.

03

Answers shift weekly

AI answers are re-generated constantly. A citation you hold today can vanish next week — monitoring is the difference between finding out now and finding out from your sales numbers.

In your scan

What Speak Local checks.

Every scan measures these — scored, prioritized, and re-checked on schedule. See the full platform for how the pillars combine.

Crawler access — robots.txt, firewalls & CDNs actually letting AI bots in
Rendering — content in raw HTML, not locked behind JavaScript AI never runs
llms.txt — present, parses, and actually says something
Citation share of voice — how often AI answers name you vs. your rivals
LOCAL SEO AEO GEO AIO
GEO questions

Asked and answered.

What’s the difference between GEO and AEO?
AEO makes you quotable; GEO gets you cited. AEO is on-site craft — schema, FAQs, answer-shaped content. GEO extends off-site: whether AI crawlers can access you, whether your entity is unambiguous, and whether generative engines actually name you in answers. You need both, which is why they’re separate pillars in the same score.
How do I know if AI can read my site?
You can’t tell from a browser — you have to fetch your pages the way the bots do. Speak Local requests your site as GPTBot, ClaudeBot, and PerplexityBot from the outside and reports exactly what each one gets back, because robots.txt saying “allowed” means nothing if a firewall drops the request.
What is llms.txt?
A plain-text file at yourdomain.com/llms.txt that describes your site for language models — who you are, what you do, and which pages matter. It’s a young standard, but it’s cheap to publish, engines are starting to read it, and it’s one of the fastest fixes on most audit lists.
The answer itself

We keep the answer word for word.

Not a score for it — the text. Your name marked inside it, every rival marked beside it, and the raw version one click below so you can check the rendering changed nothing.

ChatGPT “best roofer in Lexington KY” · 14 Sep

For roof replacement in Lexington, a few local companies come up consistently. Summit Roofing1 is frequently recommended for storm damage work and carries strong reviews across Google and BBB. Apex Exteriors2 is another established option, particularly for metal roofing. Homeowners also mention Bluegrass Home Co., though I don’t have details on their current service area.

you named rival citation confirmed the domain rival name read out of the prose only

Raw text and the model id that answered sit one click below in the app.

Four rungs, not a yes or no.

  • NamedRecommended by name in the answer.
  • CitedYour page is a source under the answer.
  • ConsultedRead, but not quoted back.
  • AbsentNot there at all.

The board reports the state held by two of the last three checks, so a one-week blip is never sold to you as a change. And no assistant offers a seed that pins the index it searched — so this is one real check, on one date, not a reproducible benchmark.