The AI Visibility Guide

Why this matters, and exactly what happens when you scan.

Search is splitting. People still type into Google — but increasingly they ask ChatGPT, Gemini, Claude, and Perplexity. The brands those AIs recommend are the ones that get found. Here's how to find out where you stand, and what changes.

01 · The shift

SEO isn't dead — it's evolving.

Google's share of search is slipping for the first time in 20 years. AI-native search tools — Perplexity, ChatGPT search, Gemini, DeepSeek — are growing fast. People like asking questions and getting one direct answer instead of clicking through ten blue links.

That answer mentions one or two brands— and most users don't scroll past that. If you're not the brand the AI cites, you're invisible in the new search economy. Optimizing for Google still matters. But optimizing for AI search is a different game, with different rules.

Google search · the old game

You rank a page for a query. Users see 10 links, click yours, land on your site. SEO = backlinks + on-page keywords + technical health.

AI search · the new game

The AI recommends a brand by namein its synthesized answer. Users don't see a list — they see the answer. You either get cited or you don't.

02 · What is AI visibility

AI visibility = how often AI assistants recommend you.

When a customer asks ChatGPT “What's the best note-taking app for teams?”, the answer mentions specific brand names. AI visibility measures:

  • Cited or not? — Does the AI mention you at all?
  • Position — If listed, where (1st, 3rd, 10th)?
  • Context — How are you described? Favorably? With a competitor disclaimer?
  • Engine coverage — Are you cited across all 4 AIs, or only one?
  • Prompt breadth — How many of your customer's likely questions do you appear in?

We roll these signals into a single 0–100 visibility score. It's the AI-search equivalent of your Google rank — except it captures something Google never did: did the search even mention you in the first place?

03 · Why you need to track it

Three things you can't see without a scan.

  • Where you're invisible.You might already rank #1 on Google for “best CRM for solopreneurs.” That doesn't mean ChatGPT mentions you. Different signals, different sources, different answer.
  • Which engines like you. Claude weights citations differently than Perplexity. ChatGPT relies on training data; Gemini does live web search. You can be visible in two and absent in two without realizing it.
  • Who's eating your share.AI answers list 2–5 brands. If yours isn't one, someone else's is. We surface exactly who's being cited instead of you — and on which prompts.
Why now

AI engines update their training and citation models constantly. A brand cited yesterday may be missing tomorrow. Weekly tracking is the only way to catch drift before it costs you customers.

04 · How a scan works

What happens when you paste a domain.

You type yourdomain.comon the homepage. Here's the exact pipeline that runs in the next ~30 seconds:

1

We figure out your brand and category

We fetch your homepage, parse the title, meta description, and structured data, and infer your industry (SaaS, e-commerce, agency, finance, etc.). No login, no tracking pixel.

~2 seconds
2

We generate realistic discovery prompts

Based on your category, we build the kinds of questions your customers actually ask AIs — 'what's the best X for Y?', 'recommend a tool that does Z', 'X vs competitor'. These aren't generic keywords; they're realistic intent queries.

instant
3

We send each prompt to 4 AI engines, in parallel

ChatGPT (via the OpenAI Responses API with web search), Gemini (with Google Search grounding), Claude, and Perplexity. We capture the full answer, all cited brands, and any source URLs.

~20 seconds
4

We parse, score, and rank

For each response, we find your brand (or its absence), extract competitor mentions, measure your position in any listed answer, and aggregate into a single visibility score. The Fix List is computed from the gaps.

~5 seconds

Total elapsed: ~30 seconds for a free one-shot scan, or runs on a scheduled cron for paid plans (weekly on Starter, daily on Growth).

05 · Your visibility score

The score is a blend of three signals.

We don't just count mentions. The visibility score (0–100) weighs:

  • Mention rate (70% weight) — out of all prompt × engine combinations, in how many were you cited?
  • Position quality (20% weight) — when cited, how high in the list? Position 1 is full credit; position 10+ is zero.
  • Sentiment context (10% weight) — were you cited favorably, neutrally, or as a counter-example?

0 – 39 · needs work

Cited rarely, or only in one engine. AI traffic is largely going to competitors. Highest upside if you act.

40 – 69 · mixed

You're cited sometimes — usually by 2 of 4 engines. Gaps to close, but a foothold exists.

70 – 100 · strong

Cited by most AIs across most prompts. Maintain the lead — competitors are catching up monthly.

06 · The Fix List

This is the part nobody else gives you.

A score on its own is just a number. The Fix List turns the score into a punch-list: every fix has a specific change, a location, a reason, and an estimated point impact. Here's a real example of one fix entry:

Add a comparison FAQ block to /pricing

+20 pts
Locationyourdomain.com/pricingWhyGemini and Perplexity surfaced competitor FAQ pages when asked “Is X worth it for small businesses?” Your /pricing page has zero question-format content.EffortMedium · ~90 minutesImpactHigh · likely +15 to +25 visibility points within 2 weeks
Suggested draft

Is AI Rank Grow worth it for small businesses? Yes — the Free plan gives you a real, full-engine scan with no signup. Paid plans (from $19/mo) are typical for serious tracking, and most users see measurable rank gains within 30 days. [3 more Q&As suggested in your full Fix List...]

Each fix falls into one of four categories:

  • On-page rewrites — exact paragraphs, FAQ blocks, comparison sections to add
  • Schema markup — JSON-LD snippets ready to paste into your <head>
  • Citation targets — specific third-party publications, directories, and roundups to pitch
  • Content gaps — pages that don't exist yet but should (vs-competitor, alternatives-to, best-for-X)
07 · What to do after

The 30-day improvement loop.

Most users follow this pattern after the first scan:

  • Week 1 — Ship the top 3 high-impact fixes (usually on-page rewrites)
  • Week 2 — Embed the schema markup, submit to 2 of the citation targets
  • Week 3 — Publish the missing comparison/alternatives pages
  • Week 4 — Re-scan. Track which engines moved. Adjust the next 30 days.

Real-world result: +15 to +30 pointsin the first month is typical when the top 3 fixes get done. The gains compound — once you're in the training data for ChatGPT and Claude, you stay there.

08 · Common questions

Anything else?

  • Will AI engines penalize me for “optimizing”?No. We don't suggest keyword stuffing or fake content. The Fix List is just good content: clearer pricing, honest FAQs, better schema. AI engines reward clarity.
  • Do I need to install anything?No script tag, no analytics integration, no DNS change. Paste a domain — that's it.
  • Is the data noisy? Yes, individual AI responses are non-deterministic. We mitigate by running multiple prompts per scan and averaging across runs. The weekly rolling average is what matters for trend.
  • What about local SEO / map pack / voice search? Out of scope for v1. We focus on chat-style AI assistants where the answer is a sentence, not a list of map pins. Voice (Alexa, Siri) is on the roadmap.

Run your first scan.

30 seconds, no signup. See your score, see your fix list, see who's being cited instead of you.

Check my visibility →