AEO Tools: What They Really Measure and What They Cost
Dominik Breitbach founded taismo GmbH, an SEO and GEO agency from Munich, and works as lead SEO strategist on SEO retainers and visibility in AI answers. He has run both tool classes in this article in daily client work for months, including the quota math that never makes it onto a product page.
AEO tools ask language models questions and record what the answers say. They measure the answers about your website, and that detail decides everything: There are two kinds of AEO tools, and they answer two different questions. One measures your own visibility against the keywords you already track. The other measures the market, meaning which brands the models name for your topics and which sources they build their answers from. Buy the wrong kind and you measure the wrong thing very precisely. We have run both in daily work for months, and this guide shows what they deliver, what they cost and where they stop.
👉 Want to know where your page stands before you buy a tool? Our free AI visibility checker gives you a first answer in two minutes.
AEO tools in short
An AEO tool does three things. It sends a list of prompts to one or several language models, it stores the answers, and it evaluates whether your brand appears, whether your site is linked and which sources the model cited. That is the whole mechanism. It is also why the choice of prompts matters more than the interface of the tool.
A note on the terms, because the market uses several of them. AEO stands for answer engine optimization, the work of earning a place in AI answers. The same discipline is also called generative engine optimization or GEO, and you will find the same tools sold as AEO tools, GEO tools, AI visibility tools or LLM tracking tools. In the US the label “GEO tools” carries extra noise: Search results for it mix AI visibility platforms with GeoTools, an open source Java toolkit for geographic data. If you research the market, “AEO tools” or “AI visibility tools” return cleaner results.
The category is young and crowded. In the US results for “AEO tools” on 4 October 2026, all ten positions went to comparison lists, product pages and a G2 software category, and a CRM vendor as large as HubSpot now sells its own AEO product. Two related entries in our glossary help you sort the vocabulary: LLMO and AI search.
Two questions, two tool classes
Comparison lists sort AEO tools by features and price. A more useful first sort is the question you want answered, because the market splits into two classes along that line.
Question 1: Are you mentioned for your topics?
This is the inside view. It runs against the keywords you already track and answers a narrow, very concrete question: Does your domain appear in the AI answers for these terms, and does that match your organic rankings?
Question 2: Who gets mentioned in your place?
This is the market view. It runs against a set of prompts you write yourself and answers: Which brands does the model name for this topic, in what proportion, in what tone, and from which sources does it build the answer?
Both questions are legitimate. They lead to different invoices, different plans and different conclusions. If you buy the inside view and expect a market analysis, you get a clean measurement that answers a question you did not ask. Our generative engine optimization services start with exactly this choice, because it decides what you pay for every month after.
Class 1: Your own visibility against tracked keywords
Tools in this class attach to your existing rank tracking project. They take the keywords that already sit there, check for each one whether an AI answer appears, and log whether your domain is mentioned or linked. Setup is quick, because the keyword list already exists.
The overlap with organic rankings is the real finding
The most useful number in this class is the overlap with your organic rankings, and it splits into four fields:
- You rank and you get cited. The normal case for pages that work. Nothing to do.
- You rank well and the AI answer skips you. The most expensive finding. The page is findable and hard to quote: The answer sits too far down, the statement is blurred, the intro runs too long. This is where the work is.
- You get cited and do not rank. More common than people expect. The page carries more weight in the answer than in the results list, often because it delivers a clean definition.
- Neither. This is a classic SEO task, and AEO starts after it.
Only the second and third fields trigger action, the rest confirms what you know. If your tool gives you one aggregated visibility score instead, you face the same compression as a classic visibility index: Convenient for the trend line and too coarse to decide what you work on tomorrow.
Class 2: The market
Tools in this class attach to prompts you write yourself. They run against several models and deliver four things:
- Share of voice: How often your brand gets named across the whole prompt set, measured against all brands named.
- Competitor brands: Which providers appear for the same prompts. Often the most sobering column, because it regularly lists names that play no role in your organic competition.
- Sentiment: The tone in which the models talk about your brand.
- Cited sources at domain and URL level: Which pages the model actually drew on.
Cited sources are the most practical number
Of these four outputs, the cited sources are the one that turns directly into work. You see which single URL earns a competitor the mention. You also see that models often cite comparison portals, trade articles and directories, and vendor pages far less often. Once you have seen that, you understand why brand mentions count even without a link, and why a place on a third-party list can bring more than one more page on your own site. The mechanics behind the selection are explained in our article on how AI systems decide which brands they recommend.
What AEO tools really cost
The price of an AEO tool sits in two quotas, and the plan line shows only part of it.
First: A prompt costs in every measurement cycle. If you run 100 prompts weekly, four weeks consume four times that. You are buying a subscription with a volume cap.
Second: A prompt counts once, no matter how many models it runs against. We checked this ourselves because no product description said it: A project with 20 prompts against five models consumed exactly 20 of 200 prompts. For your planning, that means more models cost nothing and more prompts cost everything.
The project count is the real bottleneck
We run both tool classes with the same vendor, SE Ranking, and even there they are two separate products with two separate quotas: The AI Results Tracker for the inside view with 450 prompts, and SE Visible for the market view with 200 prompts and three projects. Those are the numbers of our plan, so check them for yours. The 200 prompts look like the hard limit. The three projects limit you first. One project is one brand, so three projects let you measure three brands at the same time, and a fourth brand has to wait however many prompts remain free.
For an agency, that is the actual buying decision, and it reverses the order of the evaluation. Prompts can be allocated, projects cannot. At taismo, clients with an AEO mandate get 25 to 40 prompts, clients with an ongoing SEO retainer 12 to 15, everyone else none. Write such a tier list down before you start, otherwise the client who asked first uses up the quota in week one.
Which prompts deserve a slot in your AEO tool
When prompts are the scarce resource, choosing them is the real work. That requires a distinction that gets blurred every day.
An FAQ and a prompt do different jobs. An FAQ is a question you ask so that your own page answers it. A prompt is a question a person asks a model. One is content, the other is measurement. If you pour the FAQ collection of your website into a tracking tool, you burn the quota on questions nobody asks that way. For scale: Our own site carries between 2,400 and 6,100 FAQs across 2,445 indexed URLs, depending on how you count. No plan covers that, including the large tier with 1,000 prompts.
Our allocation rule has two conditions, and both must be true: A prompt earns a slot when the answer carries a vendor or buying decision and when it distinguishes between vendors.
That applies to selection prompts (“which X is best”), comparison prompts (“X or Y”), brand prompts (does the model name your brand or a competitor) and, for local businesses, location prompts. Definition questions without vendor context, service questions about your own operations (“how long does delivery take with you”) and questions with trivial factual answers belong on your pages. In a tool, the model names no brand for them and the slot burns. Our guide to prompt tracking walks through a complete 15 prompt plan.
Test the data route before you build reports
One point appears in no tool review, because it only surfaces once you process the data automatically: Two tools from the same vendor can require separate access keys. In our setup, the key of the rank tracking API is invalid for the market tool, and the answer is a 401 that you understand immediately.
The second case is less obvious. With the right key and without a browser user agent in the request header, the API itself never answers. The protection layer in front of it does: A 403 with its own error code. That 403 looks like a permissions problem and is a header problem. If you read it as missing rights, you search your account settings first, and that costs half a day.
The general lesson: If you plan to build a recurring report from an AEO tool, test the API route before you build the report. An interface that works in the browser tells you nothing about whether the API answers on the same terms.
Three blind spots every AEO tool shares
Google states the boundary in one sentence in its guide to optimizing for generative AI features: “No third-party tool has access to our internal ranking or AI systems.” Everything else follows from that. Every AEO tool works by asking, so it produces a sample, and three things lie outside its reach.
First, real frequencies. A tool knows what a model answered to your 200 prompts. It has no data on how often people actually ask those questions. The only place with real Google data is the generative AI performance report in Google Search Console: Free, available to all websites worldwide since 31 August 2026, covering AI Overviews and AI Mode. By design it reports impressions only, without clicks, positions or queries.
Second, the stability of a single answer. The same model answers the same prompt differently at different times. A single run is an anecdote. A series over weeks is a measurement, and that is exactly why a prompt costs in every cycle.
Third, revenue. A tool shows the mention and stops there. The path from an AI answer to your contact form leaves no continuous trace the tool could read. Our guide to measuring AI search visibility covers the sources that close part of that gap, from referral traffic to server logs.
We set up the measurement, choose the prompts and deliver the trend, so you get answers without learning another tool. Choosing the prompts is the part everything depends on.
Five questions before you buy an AEO tool
- Which of the two questions do you want answered? Inside view or market view. If you need both, you need two tools, and it pays to know that before you sign.
- How many projects does the plan include? The project count limits how many brands you can measure at once, and frugal use of prompts does nothing to raise it.
- Does the tool show cited sources at URL level? At domain level the information looks nice. At URL level it becomes a work order.
- How often does it measure, and what does one cycle cost? Calculate the month, never the single run.
- Can you reach the raw data? Export or API. Otherwise your report depends on an interface the vendor can redesign at any time.
A two-week trial tells you more than any demo
Most vendors offer a trial, and two weeks are enough to answer the five questions with your own data. Step 1: Pick 10 prompts that pass the two-condition test, including two brand prompts. Step 2: Run them in the trial tool and, on the same day, ask the same prompts manually in ChatGPT and in Google AI Mode. Step 3: Compare the brand lists. If the tool reports brands the manual check never shows, ask the vendor how it sets location and language. Step 4: Export the raw data once and open it in a spreadsheet. If the export misses the cited URLs, you know what the tool will never tell you. Step 5: Repeat the run after one week and check whether the change between both runs is larger than the variation you would expect from a probabilistic model. A tool that reports every wobble as a trend makes your monthly report louder and your decisions worse.
If you want a first assessment without a plan or an account, our free AI visibility checker tests your page for the points that decide whether a model can cite you at all.
Visibility comes from the groundwork the tool measures
An AEO tool tells you where you stand, and after two months it tells you whether anything moved. The movement itself comes from your pages.
What moves it is everything that makes a model find your content worth quoting: Clear answers placed where the question is asked, clean entities that make the same statement across pages and domains, schema markup that repeats that statement in machine-readable form, and a site that AI systems can fetch, which can include a maintained llms.txt. In the same guide, Google writes a sentence that sorts out much of the excitement around the labels: “optimizing for generative AI search is optimizing for the search experience, and thus still SEO.”
With the groundwork in place, an AEO tool becomes a precise instrument for steering. The order is therefore clear: Pages first, measurement second.
We recommend no tool plan before we know which prompts your topic actually carries. The first call is free, and it often ends with the advice to work on your pages first and on the measurement second.
Frequently asked questions about AEO tools
What is an AEO tool?
An AEO tool sends a fixed set of prompts to language models such as ChatGPT, Gemini or Perplexity and evaluates the answers: Whether a brand is named, whether it is linked and which sources the model cited. AEO stands for answer engine optimization, the same discipline that is also called generative engine optimization or GEO.
Are AEO tools and GEO tools the same?
Yes. AEO tools, GEO tools, AI visibility tools and LLM tracking tools describe the same product category. The labels differ by market and vendor, and the measurement behind them is the same.
How much does an AEO tool cost?
The plan price is only half the answer. Two quotas decide the real cost: The number of prompts, which is consumed again in every cycle, and the number of projects, meaning the brands you can measure at the same time. With weekly runs, a 200 prompt quota covers 50 prompts.
What is the best AEO tool?
The best AEO tool is the one that answers your question. To track your own visibility for the keywords you rank for, choose a tool that attaches to rank tracking. To see who gets named in your place and which sources the models cite, choose a market monitoring tool with URL level source data.
Is Google Search Console enough to measure AI visibility?
For part of the question, yes. Since 31 August 2026, the generative AI performance report shows worldwide how often links to a site appeared in AI Overviews and AI Mode. It reports impressions only, without clicks, positions or queries, and it covers Google alone.
Why do two AEO tools show different results?
They ask different questions. Results depend on the prompt set, the model, the time and the location of the query, and every tool works with samples. Two tools are only comparable when they run the same prompts at the same time.
Sources
- Google Search Central: “Google’s guide to optimizing for generative AI features”, developers.google.com, accessed 4 October 2026.
- Google Search Console Help: “Generative AI performance report (Google Search)”, support.google.com, accessed 4 October 2026.
- Google Search Central Blog: “Introducing Search Generative AI performance reports in Search Console”, developers.google.com, 3 June 2026, updated 31 August 2026.
- SE Ranking: “AI Visibility Tracker”, seranking.com, accessed 4 October 2026.
