THE THING WE BUILT:

AI Search Readiness

Nobody outside these companies fully knows how ChatGPT, Gemini, or Claude decide which companies and sources to surface.

That has not stopped a wave of “GEO scores”: proprietary numbers that often compress a messy, changing set of observations into something that looks more definitive than it is. It feels like SEO theater with an AI sticker on it.

So you’ve sign up for a trial… and you get the score: 81. Or “above average visibility.” Cool. Now what? Which page do you write? Which buyer question are you missing? The number feels finished. The work is not.

More and more, AEO and GEO come up early in a call. Clients are asking the right questions. They want to know whether their site can show up when someone asks an AI for help. Fair. The hard part is that the category is still early: standards are unsettled, tools disagree, and it is easy to treat the acronyms as a checkbox instead of a real content challenge to address.

When we run those tools on our own site, we keep wondering: are we asking the right question?

Not “did we check GEO and AEO off the list.”
Not “what is our score.”

Is this site ready for AI search?

And more specifically: ready for the conversations we want to win, not only the ones the homepage already covers.

Those conversations became the foundation of this prototype.

We started building a library of the kinds of buyer questions we hope AI can answer with confidence. Not keywords. Actual questions people ask when they're evaluating vendors, products, and partners.

The goal isn't another score. It's to show you the buyer conversations your website already supports—and, more importantly, the ones it's missing.

By that, we mean something specific:

This prototype assesses AI Search Readiness by asking one question: Does a website clearly and credibly address the buyer questions that could lead to an AI-assisted recommendation?

It doesn’t try to predict placement in ChatGPT, Gemini, Claude, or any other AI product. You can't inspect or score those ranking systems directly. What you can evaluate is whether your website contains strong material for the questions your audience is asking.

Think of it more like Lighthouse than a leaderboard score: surface the checks, expose the gaps, and point people toward what to improve next. Not official. Not a forecast. Actionable.

Your AI Search Readiness assessment

How It Works

Paste any website URL. Optionally add a few notes: markets you want to break into, how you want to be seen, offers you are pushing.

That last bit matters. Most audits only score where you already are. We also want to ask where you are going. If the live site is still thin on a market you care about, that should show up as a missing conversation, not as a silent pass.

1. Crawl the site. Sitemap when we can, otherwise homepage links. Title, meta, H1, short excerpt: enough to judge answer quality.

2. Evaluate against a buyer question library. Not keywords. Questions people actually ask AI:

• Who builds AI-integrated customer portals?

• Medical device software development company

• Who should redesign our enterprise website?

• How do I modernize a legacy app without rebuilding everything?

3. Grade each buyer question. Excellent / Good / Weak / None, plus the best matching page and a one-line reason.

4. Lead with what's missing. The useful output isn't "Your score is 72." It’s:

> You cover "Next.js agency" and "AI integrations."

> But if someone asks "Who builds AI-integrated customer portals?" your site doesn’t have a page that answers that well.

Then suggest what to do next: six concrete service or answer pages, each with a title, why it matters, and a short outline. Something you can take into a content meeting, not another audit PDF.

Notes from Building

This started as a manual exercise for the Township website.

After trying a handful of GEO tools, we found ourselves answering lots of onboarding questions upfront, waiting for a score, and then wondering... okay, now what?

So we tried our own take.

We crawled our own site, exported everything to a CSV, built a library of potential buyer questions, and started evaluating pages. With a little work, we actually ended up with something useful: a concrete list of gaps, missing buyer conversations, and pages we wanted to write.

The prototype is really just an attempt to make that process repeatable. Paste any URL. Add a little context about where you want to go. See the conversations your website already supports—and the ones it doesn’t. Leave with a content plan instead of another score.

That feels a lot closer to the question clients are actually asking.

Built with ❤️ by Township

You’re receiving this because you signed up for Township’s “We Built a Thing” newsletter. We'll only send these when we, well, build things.