Guide

AEO Tracker: How to Track Your Brand in ChatGPT, Perplexity & AI Overviews (2026)

Last updated 13 October 2026Prices checked 10 October 2026

What an AEO tracker measures

A useful AEO tracker answers four questions for every prompt:

1. Do we appear? Is the brand named in the answer?
2. Are we cited? Is our domain one of the sources the engine links to?
3. Who appears instead? Which competitors and third-party pages get named?
4. Is it moving? Is the share rising or falling, per engine?

Terradium reports these as appearance rate (the share of sampled prompts where your brand shows up at all, mentioned or cited), citation share (how often you are the named source the engine links to), share of voice (your slice of the cited sources against the competitors on the same prompts) and average position (where you land among cited sources when you do appear). Each shows a current value and a trend.

Mentions and citations are not the same number. A brand can be named in a ChatGPT answer while a competitor's comparison page is the cited source. Citations are what a page can influence most directly, so track them separately from mentions.

Why rank trackers miss AI answers

A rank tracker checks one query, in one market, at one moment, and records a position in a list of links. That model breaks in AI search:

• Answers vary run to run. The same prompt can return different sources each time, and a small wording change shifts who gets quoted. One check is a screenshot, not a measurement.
• The unit is a question. People type full questions into ChatGPT, not two-word keywords.
• There is no page one. An AI Overview cites a handful of sources. You are in that set or you are not, and a blue-link position does not guarantee a citation.

So AEO tracking uses prompt sampling: ask the same prompts repeatedly and report the rate. The AI rank tracker comparison covers the same idea across a wider set of tools.

How AEO tracking works

Every tracker is built from four settings. Compare tools on all four, not on a prompt count alone.

Prompt sets

The questions you track, grouped by topic and worded the way a buyer asks an assistant. Mix three kinds:

• Category questions, such as "best AEO tracker for a small team".
• Comparison questions, such as "Peec vs Otterly".
• Problem questions, such as "how do I get cited in AI Overviews".

Start with the unbranded questions closest to a purchase. A branded question scores high only because the brand is already in it.

Engines

The common set is ChatGPT, Perplexity, Google AI Overviews and Gemini. Some tools add Microsoft Copilot, Claude or Google AI Mode. Ask which assistants your customers use before paying for extra coverage.

Locations

Google AI Overviews are geographic. The same question returns different sources in different countries. In Terradium, the brand profile's country and language are applied to AI Overview sampling: the country selects the Google market, the language selects the answer language. Chat engines take no market. They answer in the language of the question, which is what a person typing that question would see.

Cadence

Daily sampling settles a trend faster. Weekly sampling costs less and is usually enough to see whether a new page is getting cited, because citations move on the scale of weeks, not hours. A daily plan samples roughly seven times as many answers as a weekly one for the same prompt count, so compare answers sampled per month (prompts × runs × engines), not the prompt number on the pricing page.

Metrics that matter

The test for a metric: does it tell you what to write or fix next? If it does not, it is a report, not a tracker.

Terradium, Peec, Profound and Otterly

Four tools that show up when people look for an AEO tracker. Every price below links to the vendor's own pricing page and was checked on the date at the top of this page. Entry plans, billed monthly. Where a vendor does not publish a number, the cell says so.

A wider set, including Semrush, Ahrefs Brand Radar, Scrunch, AthenaHQ and Rankscale, is on the best AI visibility tools page.

Profound's paid plan does not publish a prompt count or a price. The 50 prompts in its free-trial cell are the 7-day trial, run daily on ChatGPT, Gemini and Google AI Overviews, not an Enterprise allowance. Enterprise lists up to 9 answer engines, at a custom price.

Peec AI Starter is $95 a month for 50 prompts tracked daily on 3 engines of your choice ($80 billed annually). Otterly.AI Lite is $29 a month for 15 prompts tracked daily on ChatGPT, Google AI Overviews, Perplexity and Microsoft Copilot. Google AI Mode, Gemini and Claude are paid add-ons on Otterly. Terradium Starter is $29 a month for 15 questions, weekly, on 3 of ChatGPT, Perplexity, Google AI Overviews and Gemini, with 30 article generations in the period.

How to read it. Peec and Otterly sample daily, so they record more answers per prompt than Terradium's weekly plan. Profound is the widest engine list, on a sales contract, with no published paid price. Terradium samples less often and includes 30 articles a month aimed at the gaps. The next published step up is Peec Pro at $245 for 150 prompts, or Otterly Standard at $189 for 100. Daily sampling and 50 questions on Terradium are a higher plan that is not publicly listed.

Writing and publishing are not on Peec, Profound or Otterly's pricing pages, so they are not in the table. Terradium's $29 plan includes 30 article generations a billing period and publishes to a built-in CMS or a signed webhook.

Cost per 100 prompts

The same prompt count is not the same amount of measurement. This is the entry plan's monthly price against the prompts it tracks, and against the answers it actually samples.

Per 100 prompts = monthly price ÷ prompts × 100. Answers sampled = prompts × runs per month (daily 30.4, weekly 4.35) × engines included. Per 100 answers = monthly price ÷ answers × 100. Entry plans, billed monthly, as listed on each vendor's pricing page.

Which tracker fits you

Quick setup in Terradium

This is Terradium's onboarding flow. A checklist in the dashboard ticks as each step lands. Keyword research needs the 7-day trial.

1. Create a project. Projects, then New project. Add a name, website URL, product context and an optional image style. See Projects.
2. Brand profile. What you sell, who you sell it to, and the country and language your customers search in. AI Overview sampling uses that country and language. See AI Visibility.
3. Brand and entity. Organization name, logo, sameAs profile links and a default author. These fill the JSON-LD graph on every published post.
4. Competitors. The domains you lose customers to. Three to five is enough to start.
5. Keyword research. Confirm 10 to 20 seed keywords, short commercial phrases a customer would search. Terradium clusters the results. Confirm the list before the job runs: it spends a keyword-research run.
6. Topics and prompts. Turn the clusters into 3 to 6 topics and the questions buyers actually ask an assistant. The $29 plan tracks 15 of those questions.
7. Sampling schedule. Under AI Visibility, pick your 3 engines. The card projects questions × engines × runs × cycles against the period's allowance.
8. First content plan. Terradium drafts a plan from the clusters. It stays a draft until you approve it in Plans. Nothing is generated or published before that.

Using a coding agent? The MCP setup guide walks the same steps and stops at a drafted plan.

From tracking to action

Tracking pays when it changes what you publish. In Terradium that list is Actions, derived from the latest scan each time you open it. There is no separate actions table.

1. Find the gaps. Prompts are ranked by weakness: never appear, appear but are not cited, or cited low. Each row names the engines that did not cite you, whether a tracked competitor was cited, and which competitor URLs were. Those URLs come from the citations stored with the sample. Terradium does not crawl the competitor's site.
2. Pick the format. Each action suggests a listicle for "best X", a comparison with a real table for "X vs Y", or a how-to with ordered steps for "how do I". You can change the suggestion before generating.
3. Generate. Write it runs a draft aimed at that gap, counted against 30 generations per period. The article stays linked to the prompt, so the next scans show whether the citation rate moved. An editorial plan schedules a calendar, one article per cluster, and stays a draft until you approve it.
4. Check readiness. Most AI crawlers do not run JavaScript. GEO Readiness fetches your live site and grades setup, the published page (raw HTML, JSON-LD, canonical), discoverability files (robots.txt, sitemap.xml, llms.txt) and crawler access. A blocking failure holds auto-publish until it is fixed. The free AI crawler checker is the same idea for a single URL, with no account.
5. Publish. To the built-in Terradium CMS, served by the content API, or to a signed webhook. Auto-publish waits for a review window, 24 hours by default. If the site is statically built, connect the repository and Terradium triggers a GitHub Actions build when an article goes live. The articles are not committed into the repo. The build pulls them from the content API. See Publishing.
6. Re-measure. The next weekly scan shows whether the page is being cited. Gap, plan, draft, publish, measure: that loop is what separates an AEO tracker from a dashboard.

A public example from kugie.app: Acsent, an AI IELTS and TOEFL prep brand, is cited in 15% of 179 sampled answers over 28 days in Terradium AI Visibility.

Questions

What is the difference between AEO and GEO?

Both mean getting your brand used in AI-generated answers. AEO usually refers to answer surfaces such as Google AI Overviews. GEO is the broader term for generative search. The tracking and the content work are the same.

How many prompts should I track?

Start with the questions closest to a purchase: category, comparison and problem prompts for your main product. Fifteen well-chosen prompts, tracked the same way each cycle, beat a long list of generic ones. Terradium's $29 plan tracks 15.

Daily or weekly?

Daily sampling produces more answers, so a trend settles faster. Weekly sampling costs less and is usually enough to see whether new content is getting cited. A daily plan samples about seven times as many answers as a weekly one, so compare answers sampled per month, not prompt counts.

Can a tracker improve visibility by itself?

No. A tracker only measures. You need pages AI engines can find, read and quote, on a site their crawlers can access. Terradium pairs the measurement with an AI writer: 30 articles a month on the $29 plan. The free AI crawler checker shows whether crawlers can read your site.

Which engines does Terradium track?

ChatGPT, Perplexity, Google AI Overviews and Gemini. The $29 plan samples three of those four, chosen by you, once a week, one run per question.

Is there a free AEO tracker?

For a one-off check, yes. Terradium's free AI visibility checker asks ChatGPT and Google AI Overviews one question and reports whether each answer mentions your brand and cites your domain. One question needs no sign-up. An email unlocks up to 10 questions and a weekly re-check.

Start tracking, then start closing gaps

Run a free sample prompt check. It asks ChatGPT and Google AI Overviews one question and reports whether each answer mentions your brand and cites your domain. No sign-up for one question.

Or start free for 7 days and get a gap list and a content plan from the same tool.