HomeWhat we fixAI SEO & GEO

AI SEO & GEO

AI is answering your customers. It doesn't know you exist.

Your buyers stopped typing keywords into Google. They ask AI assistants which company to use. They ask AI search engines what to look for. They read the AI summary at the top of the search results and never scroll past it. In each of those moments, a handful of businesses get named — and if you're not one of them, you weren't in the conversation at all. Not ranked lower. Absent.

14 years in SEOUS businessesNo guarantees we can't keep

Find out what AI says about you

We'll check whether AI systems can read your site, whether they understand who you are, and what they say when someone asks about your category.

Free · About a minute · No account needed

Why scan before buying? Because most sites have one specific blocker rather than a general problem — and until we look, neither of us knows which. The scan tells both of us, and you keep the report either way.
What we find

What usually goes wrong with AI visibility

Every item below is something our scanner genuinely checks. Run it and you'll see which of these are on your site — in your own report, in these exact words. Most sites have three or four. The first one on this list is on roughly a third of the sites we scan, and almost nobody knows.

You're blocking the crawlers

In 2023, a lot of sites added rules to robots.txt to keep AI companies from training on their content. It was a reasonable reaction at the time. Most of those sites never revisited the decision.

But those same crawlers are now how AI assistants find out what exists. Blocking GPTBot doesn't stop your competitor being recommended — it stops you being available to recommend. The rule that was meant to protect your content is now the reason nobody sees it.

Worse, a lot of these blocks were never deliberate. A developer copied a robots.txt template from somewhere, or an SEO plugin added rules by default. The site owner has no idea it happened.

What this looks like

A professional services firm couldn't understand why they never appeared in AI answers despite ranking well in Google. Their robots.txt had four Disallow lines added by a plugin two years earlier, blocking GPTBot, ClaudeBot and two others. Nobody had looked at that file since it was created.

Checked by: AI Crawler Access · Robots.txt Blocking Crawlers

Nothing tells AI who you actually are

An AI system reading your site has to work out what your business is, where it operates, and what it does — from prose written for humans. Sometimes it gets it right. Often it gets it approximately right, which is worse, because approximately right becomes confidently wrong when it repeats the answer.

Structured data solves this. Organization or LocalBusiness schema states your name, location, category and contact details in a form that leaves no room for interpretation. Most sites don't have it, and the ones that do often have it half-filled or invalid.

The symptom is subtle: AI describes you nearly correctly. Wrong city. Wrong category. Confused with a similarly-named company. You never see it happen, because it happens in a conversation you're not part of.

What this looks like

A regional accounting firm found that when asked about them, an AI assistant described them as a bookkeeping service in a neighbouring city. Both details were wrong. Their site said the right things in paragraphs; it had no structured data saying them unambiguously.

Checked by: Entity Clarity · Structured Data (Schema)

Your answers are buried three paragraphs down

Human readers scroll. AI systems extract. When an AI needs a specific answer, it looks for content that states it directly and early — and moves on if it doesn't find one.

Most business writing does the opposite. It builds up. There's context first, then background, then finally the answer somewhere in paragraph four. That reads fine to a person who's already interested. To a system scanning for extractable answers, it looks like a page that doesn't answer the question.

This is the difference between content that's good and content that's quotable. They're not the same thing, and most sites have the first without the second.

What this looks like

A clinic's page on a specific treatment ran 1,400 words and was genuinely well written. But the direct answer to "how long does recovery take" appeared in the eleventh paragraph, phrased conversationally. A competitor with a thinner page had it in a single sentence under a clear heading. Guess which one got quoted.

Checked by: Direct-Answer Structure · Content Depth / Word Count

The questions your customers ask aren't marked up

Every business gets the same handful of questions before someone buys. How much. How long. Do you work with X. What happens if it doesn't work.

Those questions are the most extractable content you can publish — they're literally in question-and-answer form, which is the format AI systems handle best. But they only work if they're actually on the site and marked up as FAQs. Answering them well on a phone call helps nobody who's asking an AI at eleven at night.

Two failures here, and they're different: some sites have no FAQ content at all, and some have good FAQ content that isn't marked up, so it reads as ordinary prose to a machine.

What this looks like

A home services company answered the same six questions on every sales call. None of them appeared anywhere on the website. Their competitor had a FAQ page with schema markup, and consistently showed up in AI answers for exactly those questions — using answers that were, honestly, worse.

Checked by: FAQ Schema Markup · Visible FAQ Content

Nothing says who wrote this or why they'd know

AI systems weigh credibility, and credibility signals are mostly absent from small business websites. No author. No credentials. No indication whether this was written by someone with twenty years of experience or generated last Tuesday.

This matters more now than it did, because there's far more content than there used to be and a lot of it is worthless. Systems that have to choose what to trust look for signals of who stands behind a claim. A page with no author attached is a page with nobody standing behind it.

The same applies to freshness. Content with no date, never updated, reads as possibly-stale — and a system deciding between two answers will prefer the one that looks current.

What this looks like

Two sites covered the same topic. One had an author byline, a short bio establishing relevant experience, and a visible last-updated date. The other had neither. The first was cited noticeably more often, despite the second being longer and arguably more thorough.

Checked by: Author / E-E-A-T Signals · Content Freshness Signals

Your details don't match anywhere else

Your website says one thing. Your Google Business Profile says something slightly different. An old directory listing has a phone number you stopped using in 2021. Each is close, none matches exactly.

Humans don't notice. Systems do — and inconsistency reads as uncertainty. When an AI can't reconcile conflicting information about a business, the safest thing it can do is not mention that business at all. You don't get a wrong answer; you get no answer.

This is unglamorous work and it's genuinely one of the highest-return things a local business can fix. Nothing about it is technically difficult. It's just tedious, so nobody does it.

What this looks like

A local business had their address written four different ways across their site, their Google profile, and two directories — suite number in some, not others, abbreviated differently each time. After standardising all four to match exactly, the improvement in how consistently they were described was immediate.

Checked by: NAP Consistency · Contact Info Clarity · Trust / Credibility Elements

How we fix it

Found is only half the job

Other tools stop at the list. Here's exactly what we do about each one — including the steps, so you can judge whether we know what we're doing.

What we foundHow we fix it

AI crawlers blocked

Your robots.txt is keeping the systems answering your customers from reading your content.

We open access to the crawlers that matter

Line by line, against the current AI user-agent list — then verified with our own crawler checker rather than assumed.

Same dayEvery package
See exactly how
  1. Fetch and read your robots.txt line by line against the current list of AI user-agents
  2. Identify what's blocked deliberately versus what a plugin or template added without anyone deciding
  3. Open access for the crawlers you want reading you, keep out the ones you don't
  4. Re-test each user-agent individually with our checker to confirm it can now actually fetch
  5. Resubmit your sitemap so they find everything, not just whatever they stumble into

This is usually a one-line change and often the single highest-impact fix on the list. It's also the one we most often find nobody knew about.

No entity clarity

Nothing states unambiguously who you are, where you are, or what you do.

We build and validate your Organization schema

Real name, address, phone, category and profiles — written correctly and validated, not just present in the source.

2–3 daysEvery package
See exactly how
  1. Write Organization or LocalBusiness schema with your real details — not a template with placeholders left in
  2. Include the specifics that disambiguate you: exact category, service area, founding date, linked profiles
  3. Validate through Google's testing tools per template type, not once on the homepage
  4. Cross-check every detail against your Google Business Profile so nothing contradicts
  5. Re-verify after a week, because schema that validates on day one can break when a theme updates

Most of the work here isn't writing the schema. It's making the same details true in every place they appear.

Answers buried in the page

Content that builds to its point instead of stating it.

We restructure so the answer comes first

Same information, reordered so the extractable answer is at the top and the supporting detail follows.

1–2 weeksGrowth & above
See exactly how
  1. Identify the specific question each page is genuinely answering — often it's not the one the heading suggests
  2. Move the direct answer into the first two sentences under a clear heading
  3. Keep the depth beneath it — this is restructuring, not cutting
  4. Add sub-headings phrased as the questions people actually ask
  5. Send you the restructured versions to approve before anything is published

We don't shorten your content. Thin pages don't rank and don't get cited. This is about order, not length.

No FAQ content or markup

The questions your buyers ask before purchasing aren't on the site, or aren't marked up.

We write the FAQs and mark them up

Drawn from what your customers actually ask, with FAQPage schema so they can be extracted.

3–5 daysEvery package
See exactly how
  1. Collect the real questions — from your sales calls, your inbox, and what people search around your category
  2. Write answers that are genuinely useful rather than deflections toward a contact form
  3. Publish them where they belong, on relevant pages, not exiled to one FAQ page nobody reads
  4. Add FAQPage schema and validate it
  5. Add to the set over time as new questions come up

The answers have to be real answers. An FAQ that says "contact us to find out" is worse than no FAQ, because it teaches the system your page doesn't answer things.

No author or freshness signals

Nothing indicates who wrote this, why they'd know, or when it was last true.

We add real attribution and dating

Author bylines with relevant credentials, visible update dates, and Person schema where it applies.

2–3 daysGrowth & above
See exactly how
  1. Establish who should be credited — usually you, sometimes a named specialist
  2. Write a short bio that establishes genuine relevant experience, not a list of adjectives
  3. Add bylines and Person schema linking author to content
  4. Add visible last-updated dates and set a realistic review cadence
  5. Link author profiles to real professional presences where they exist

We won't invent credentials or create author personas. If the honest answer is "the founder wrote this and has done it for fourteen years," that's a stronger signal than a fabricated title.

Inconsistent business details

Your name, address or phone differs between your site, your Google profile and directories.

We make every source match exactly

Audited, standardised, and corrected wherever it appears — not just on the website.

1 weekEvery package
See exactly how
  1. Audit every place your details appear — site, Google Business Profile, directories, social profiles
  2. Decide the canonical version, down to abbreviation and suite number formatting
  3. Correct it everywhere, including the listings you forgot existed
  4. Add LocalBusiness schema stating it in machine-readable form
  5. Document what we found so you know where the inconsistencies were

This is the least interesting work on the list and one of the most reliably effective, particularly for local businesses.

Before you try this yourself

Common mistakes

What we usually find when someone has already tried to fix this themselves.

Blocking AI crawlers and forgetting

Blocking GPTBot in 2023 was a defensible call. Leaving it blocked now, without revisiting whether that's still what you want, means opting out of the fastest-growing way people find businesses. Make it a decision rather than an inheritance.

Adding schema from a generator and never validating it

Schema generators produce valid-looking output that frequently doesn't render — wrong nesting, placeholder values left in, missing required fields. Present in the source and actually working are different things, and only one of them helps.

Generating FAQ content with AI and publishing it unread

It's tempting because the format is formulaic. But AI-written answers to AI-facing questions tend toward generic, and generic doesn't get cited — the system picks whichever source says something specific. Write the answers you'd actually give on a call.

Assuming this is just SEO with a new label

It shares a foundation — a broken site fails at both. But the emphasis differs. Traditional SEO asks how you rank for a query. This asks whether you're mentioned in an answer, which depends more on clarity, structure and consistency than on position.

Our Process

Scan. See. Fixed.

01

We check what AI can see

We fetch your site the way an AI crawler does, check what's blocked, and test what systems actually say about you when asked. Free, no account needed, and you keep the report.

About a minute

02

You see exactly what's wrong

Every issue ranked by impact and effort, in plain English. Which crawlers are blocked, what's missing, and what each one is costing you.

Yours to keep, free

03

We fix it

You approve the plan before anything is touched. We do the work — crawler access, schema, content structure, consistency. You don't need a developer.

From $50/month

04

We check again and show you

The same checks, run again. You see what cleared and what changed. We also re-ask the same questions to the same systems, so you can see whether the answers moved.

Proof, not promises

The first pass is the big one — unblock, structure, mark up. After that the work is keeping you cited as your content, your competitors and the AI engines themselves keep changing. No bloated retainer, no lock-in: a monthly plan that reflects real ongoing work, and you can cancel any month.

What's included

What's included

Packages differ in how much work we do — not in how much of the picture you get.

The full scan

  • Every check across four areas
  • AI crawler access tested per user-agent
  • Entity and schema audit
  • Direct-answer structure review
  • FAQ schema and visible-FAQ audit
  • Google Business Profile lookup

The work itself

  • Crawler access opened and verified
  • Organization and LocalBusiness schema written and validated
  • Content restructured for extraction
  • FAQs written and marked up
  • Author and freshness signals added
  • Business details standardised everywhere

The proof

  • Before-and-after score comparison
  • The same questions re-asked to the same AI systems
  • Exactly which issues cleared
  • Full dashboard access
  • Every tool unlocked
  • Scan history kept permanently

What does it cost?

It depends on what your site actually needs — which is why we scan first. Plans start at $50/month, and every one includes the full scan above.

FAQ

Before you scan

Find out what AI says about you

Every check, your off-page signals, and a plan you can act on. Free, about a minute, no account needed.

Scan My Site Free