AI Visibility

Get recommended by ChatGPT, Gemini and Google

LogNorm asks the real ChatGPT, Gemini and Google AI Overviews your buyers' questions, in your country, and shows who gets mentioned, cited and recommended. Your agent keeps the prompts you track sharp and turns every learning into a fix or a page.

Start free Read the AI visibility docs
Works withClaude CodeClaude CodeCodexCodexCursorCursorClaudeClaudeand any MCP client
AI Visibility · acme.com · asked from the United States
Tracked prompt
“Best invoicing software for a five-person design studio”
CategoryConsiderationBuyer intent
EngineMentions AcmeCites
CChatGPTRecommendedCited
GGeminiNot mentionedCited
AGoogle AI OverviewsTop pickNot cited
Mention rate
50%
2 of 4 answers name you
Citation rate
50%
2 of 4 cite your pages

Illustrative example. Acme is a sample brand.

Live in your workspace
  • Panda · Pip turned a learning into Answer-first intro for /pricing
    A ranked move
  • Sam's Panda added prompts best project management software for startups
    3 new buyer questions tracked
  • LogNorm collected 24 AI answers
    ChatGPT, Gemini and Google AI Overviews
Why it matters

Your buyers ask AI first. Are you the answer?

01

AI assistants recommend your competitors and you never see it.

LogNorm asks the engines' real products your buyers' questions and records every mention, citation and recommendation.
02

Knowing you're missing doesn't tell you what to do.

Ranked learnings name the page to change or the source to get listed on, and become moves.
03

Tracking the wrong prompts measures the wrong thing.

Your agent keeps the prompt list matched to real buyer questions: it adds, edits, pauses and retires them.
Humans and agents, in parallel

From not mentioned to recommended

You and your team
Decide, review, publish
  1. 01Pick the buyer questions that matter
  2. 02Approve page changes and outreach
  3. 03Watch your share of recommendations
Your agents
Claude Code, Codex, Cursor, Claude
Claude CodeCodexCursorClaude
  1. 01Add, edit, pause and retire tracked prompts
  2. 02Turn learnings into moves and GEO fixes in your repo
  3. 03Draft answer-first pages the engines can quote
LogNorm
Finds, ranks, verifies, measures
  1. 01Ask ChatGPT, Gemini and AI Overviews on schedule
  2. 02Separate mentions, citations and recommendations
  3. 03Find the sources that cite competitors, not you
Prompts

Start from the questions your buyers actually ask

LogNorm suggests prompts from your positioning, your topics and your competitors, phrased the way someone asks before they have heard of you. Pick the ones worth tracking, or write your own.
  • Every prompt is classified: category, alternatives, how-to, comparison or evaluation.
  • Funnel stage and buyer intent, so you know which prompts are close to a purchase.
  • Asked in the engines' real products, from your market's country.
Suggested prompts
Suggestions for acme.com
Written from your positioning, topics and competitors. Your name never appears in them.
  • Best invoicing software for freelancers who bill in several currencies
    Category91%ConsiderationTopic: Invoicing
  • Globex alternatives for small agencies
    Alternatives94%DecisionTopic: Switching
  • How do I chase late client payments without being awkward
    How to72%AwarenessTopic: Payments
  • Globex vs Initech for recurring invoices
    Comparison55%DecisionNo topic yet
3 selected · asked in ChatGPT, Gemini and Google AI OverviewsTrack 3 prompts
Mentions and citations

Two facts per answer, never blended into one score

An answer can name you and cite someone else, or lean on your page without naming you. Each case calls for different work, so LogNorm keeps all four apart for you and every competitor. The four presence cases, explained.
  • Mention rate, citation rate and recommendation rate, each with its sample size.
  • Mention types show how the answer treats you: top pick, recommended, alternative, negative.
  • Your product names, former names and owned sites all count as you.
Presence · latest run, 96 answers
Acme (you)42% mentioned · 25% cited
Globex71% mentioned · 42% cited
Initech25% mentioned · 8% cited
Mentioned and cited Mentioned, not cited Cited, not mentioned Absent
Acme's pages are a source in 11% of answers that never name Acme. One blended “visibility” score would hide that.
Sources

See every page the engines cite, and the ones you are missing from

LogNorm records each URL an engine cited, linked inline or searched and left out. It classifies the sites, then reads the most-cited third-party pages to check whether they mention you.
  • Source mix by kind: your sites, competitors, review sites, publishers, forums, docs, directories, video and social.
  • Domains and pages with cited, linked and searched-only counts, prompts and engines.
  • A list of cited pages that name your competitors and not you, each one a move away from your plan.
Sources · last 4 runs
Share of answers citing each kind of source
  • Publishers64%
  • Review sites42%
  • Competitors40%
  • Forums30%
  • Your sites13%
You're missing from these
The best invoicing software for small agencies
smallbizreview.example/best-invoicing-software
14 answers cite it · Publisher · Listicle
NamesGlobexInitechDoesn't mention you
Create move
Competitors

Follow your competitors through the same answers

Every tracked competitor gets the same measurement you do, run after run. When one of them is doing something you are not, you see what: the prompts they win, the pages engines cite for them and the third-party pages that list them without you.
  • A leaderboard of mention, citation and recommendation rates with the change since the last run.
  • Where each competitor is strongest, engine by engine.
  • Brands that keep showing up untracked, ready to track in one click.
Competitors · latest run
BrandMentionedCitedRec.
Acme (you)42%25%21%
Globex71%42%54%
Prompts Globex wins
  • “Best invoicing software for a five-person design studio”Globex Top pick youAbsent
  • “Invoicing tools with retainer billing”Globex Recommended youCited, not mentioned
Initech shows up in 9 answers and isn't tracked
Track as competitor
Learnings

Ranked learnings, with the evidence attached

After each run LogNorm turns the numbers into specific findings: a listicle to get onto, a page type a competitor has and you lack, an engine that ignores you, a page of yours that is cited without earning the mention. Each carries its prompts, pages and engines, and becomes a move in one click.
  • Ranked by how much evidence supports them.
  • Sharper with every run, because changes between runs count as evidence.
  • What's working is listed too, so you keep doing it.
Learnings · ranked
  • Get listed Create move
    A listicle cited in 14 answers names Globex and Initech, not you
    Ask the editor to add Acme, with the retainer-billing angle the page lacks.
  • Cited, not named Create move
    Your pricing guide is a source in 11 answers that never name Acme
    State what Acme is and who it is for in the first screen of the guide.
  • What's working
    Your comparison page is now cited by ChatGPT and Gemini
    Keep its feature table current.
Impact

Measure whether the work you shipped changed the answers

Every page you publish and move you complete is linked to the prompts it was meant to affect. LogNorm compares those prompts in runs before and after the ship date, and tracks whether engines begin citing the new page.
  • Before and after for mention, citation and recommendation rates, with sample sizes.
  • Which engines cite the shipped page, and since when.
  • Small samples are labelled, so a hint is never presented as a result.
Impact
Acme vs Globex for agencies
Published page · shipped Sep 10 · acme.com/compare/globex
Improved
Mentioned
25%47%
22 pts
Cited
6%22%
16 pts
Recommended
13%31%
18 pts
Before: 16 answers · After: 32 answersLinked to 4 promptsCited in 7 answers, first on Sep 15
Also included

The details that make the numbers trustworthy

The real products

Answers come from the assistants your buyers use, with their own web search and sources, not from a developer API.

Your market's country

Prompts are asked from the country your website targets, so sources and recommendations match what buyers there see.

Cited, linked, searched

Three ways an engine can use a page, recorded separately for every URL in every answer.

Scheduled monitoring

Active prompts re-run daily, twice a week, weekly or monthly, and every run lands in a day-by-day history you can export.

Brand identity

Other product names and owned sites count as you in mentions and citations. Saving re-checks stored answers.

CSV export

Download every cited domain or page for the range you are viewing.
How it works

From a list of prompts to proof

Setup takes a few minutes. The AI visibility guide walks through each step.

  1. 1

    Pick prompts

    Accept suggested prompts or write your own, and choose the engines for each.
  2. 2

    Run them

    LogNorm asks every engine from your market's country and stores each answer with all of its sources.
  3. 3

    Read the learnings

    Mentions, citations, sources and competitors, distilled into ranked learnings you can turn into moves.
  4. 4

    Ship and measure

    Publish, and the Impact tab compares the answers before and after.

AI visibility questions, answered

Which AI assistants does LogNorm track?

ChatGPT, Gemini and Google AI Overviews. LogNorm asks your prompts in the products themselves rather than through developer APIs, because the products search the web, choose sources and phrase answers differently from the raw models. You choose the engines per prompt.

What is the difference between a mention and a citation?

A mention means the answer names your brand in its text. A citation means one of your pages is among the answer's sources. They are independent: an answer can name you while citing only other sites, or use your page as a source without ever naming you. LogNorm reports them separately and shows which of the four combinations each answer falls into.

Are answers the same in every country?

No. Assistants search the web from where the question is asked, so sources and recommendations change by country. LogNorm asks from the country of your website's market, so you see what a buyer there sees.

How does LogNorm decide whether an answer recommends me?

LogNorm's decision AI reads each answer and labels how it treats you and each competitor: top pick, recommended, neutral mention, alternative, negative or not mentioned. Every label shows how sure LogNorm is, and you can read the full answer behind it.

How do I know whether my work changed anything?

The Impact tab links every page you publish and every move you complete to the prompts it was meant to affect, then compares mention, citation and recommendation rates in runs before and after the ship date, with the number of answers on each side. It also shows whether engines started citing the page itself. Comparisons built on few answers are labelled as small samples.

What does a run cost?

Each delivered answer costs half a credit, plus a smaller amount for reading the answers and their sources. Twenty prompts on three engines is 60 answers, or 30 credits. Answers an engine fails to deliver cost nothing, and you see the cost of a run before you start it.

How often are prompts re-run?

Once a website has run its prompts, active prompts re-run on the schedule you pick: daily, twice a week, weekly or monthly. You can also run them any time, and the Historical runs tab shows every prompt's result day by day.

Can my agent manage AI visibility?

Yes. A connected agent reads every answer, learning and cited source, adds and retires the prompts you track (buyer questions that never name your brand), turns learnings into moves, and fixes GEO issues such as llms.txt and AI crawler access in your codebase. It can also run a check now when the work needs fresh answers.

Works with the rest of LogNorm

Growth Plan Every signal becomes a Move, ranked against the rest. Plan your week in minutes.Keywords & Topics Every keyword rated for fit, intent and format, grouped into pages and topics.Competitor Intelligence Find real competitors from search data, see their pages and every gap you can close.Site & GEO Audit A verifiable SEO audit plus a GEO audit for AI crawlers, answerability and agents.Content Studio Research-backed briefs, drafts with a live scorecard, claims checked, published to your CMS.Company Brain Your positioning, docs and customer language, used by every brief and draft.Agents (MCP) Your own coding agent claims moves, fixes issues in your repo and drafts posts, live on the dashboard.

Find out what AI assistants say about you.

Create a workspace, add a few prompts and see your mentions, citations and sources after the first run.

Start free Book a demo
Already use an agent?

Paste this into Claude Code, Codex, Cursor or Claude. It connects itself; you click Allow once.

Claude CodeClaude CodeCodexCodexCursorCursorClaudeClaudeand any MCP client