A rank tracker vs an AI visibility tool: where the questions come from, and why a keyword list does not convert

If you already pay for a rank tracker, the honest answer in October 2026 is that it probably does track AI mentions. Ahrefs sells Brand Radar, SE Ranking sells an AI visibility tracker, and both were live and readable on 20 October 2026. The argument that you need a second product because your suite ignores AI answers has expired. What decides the purchase now is narrower and more useful: where the questions come from. A rank tracker is built on a keyword database, and an AI reading is only as good as the list of questions somebody puts to the engine. Those two lists are not the same list, and one of them cannot be derived from the other.

What the big platforms actually shipped

Every row here was read on 20 October 2026, with HTTP 200 to a Chrome on Windows agent and to a Safari on macOS agent, and the response bodies saved. We make one of these products, so read the table for the facts and not for the ordering.

Read 20 Oct 2026	What it is	Surfaces named on its own page	Where its questions come from
Ahrefs Brand Radar	A module inside the Ahrefs platform	AI Overviews, Gemini, Copilot, Perplexity, ChatGPT	An index its page describes as 492 million prompts, every one "modeled from real user searches in Ahrefs' keyword database", plus a Custom Prompts feature you fill yourself
SE Ranking AI visibility tracker	A module inside the SE Ranking platform	Google AI Overviews, Google AI Mode, ChatGPT, Gemini	Your own target queries, with the prompts and citations behind each answer shown, and an API plus an MCP endpoint for pulling the data out
AI Knows Us, ours	A single purpose reading	ChatGPT, Gemini, Perplexity, Claude, Google AI Overviews	A frozen question file you write, with your brand name absent from every question

Two of the three have a large index behind them and ours does not. That is a real advantage for a category question with searchable demand behind it, and it is the reason the next section exists rather than an argument against it.

Why a keyword list does not convert into a prompt list

A keyword database is a record of what people type into a search box. An assistant gets asked things people would never type, because typing is expensive and asking is cheap. Nobody types a full sentence with their company size, their stack and their budget into Google. They do say it to ChatGPT.

We can put a number on the gap, from our own work rather than from an opinion. Building the recommendation bank for this site, we took 91 of the buying questions we wanted to measure and checked each one against Google's own US volume data. Fifteen returned any figure at all. The largest of the fifteen was a single vendor pricing query at 320 a month. Seventy six came back with no answer, which is the tool reporting that the phrase has no measurable search demand to report.

So a prompt set modelled from a keyword database starts with a structural blind spot in exactly the questions a buyer asks an assistant and never asks a search engine. The fix is not complicated and it is not automatic: write the questions yourself, from sales calls and from the words in your enquiry inbox, and put them in a file you own.

Where a rank tracker platform beats us, said plainly

• Scale of the question corpus. An index of 492 million modelled prompts will find category questions we would never think to write. Our question files are hand built and run to tens of questions, not millions.
• Data out of the building. SE Ranking publishes both an API and an MCP endpoint, so the numbers can be queried from inside another assistant. Our top plan lists a visibility API and we document no MCP endpoint.
• One invoice, one login. If the AI reading sits in the same account as your backlinks, your organic keywords and your competitors' traffic estimates, the monthly report assembles itself. Ours is a separate login and a separate line on the budget.
• Named customer references. Both platforms publish customers by name, job title and employer. Our evidence is readings and exports, which is a different kind of proof and does not replace a reference call in your own industry.

And the measurement that frames all of it: across 1,260 saved ChatGPT and Gemini answers, rival tools of every kind were named 1,314 times and our own name was named zero times. The readings behind that, including the ones that went against us, are on our proof page.

What carries over from your rank tracking setup, and what has to be rebuilt

What you already have	Does it carry over	What to do with it
Your tracked keyword list	Partly	Keep the ones that describe a buying situation. Drop the single word head terms, which an assistant answers from training with no sources attached
Your competitor list	Yes	It becomes the list of names you count in each answer, which is what turns a mention count into a share
Your location settings	Rebuild	A tracker sends a location to a search engine. An assistant is told a place or it guesses one, and the two are not interchangeable
Your position history	No	There is no position in a written answer, so the metric that replaces it is how often you are named across repeated runs
Your reporting template	Partly	Add the question count, the engine name and the date beside every rate, because an AI figure without those three is not checkable

The order worth doing it in

1. Open the AI module you already pay for and read what question list it is running. If the answer is "ours, modelled from search data", you now know the shape of its blind spot.
2. Write twenty of your own questions, in your buyers' words, with your company name in none of them. This is an afternoon and it is the part no index can do for you.
3. Run them by hand on the engines your buyers use, three times each, and label the answers yourself. That is your ground truth and it costs nothing in software.
4. Then decide what to buy, against a baseline instead of against a brochure. Sizing that decision on the smallest useful budget is covered in the smallest AI visibility budget that changes anything.

On price, we record rather than restate. We publish no figures for other vendors here, which is a position two of our own pages already state and defend. What is worth writing down from each vendor's own page is the limit that sets the tier, not the sticker: Ahrefs states that each prompt uses one check per platform, per location and per update, which means the same twenty questions cost four times as much to watch across four engines as across one.

Questions about running a rank tracker and an AI reading side by side

Does my SEO tool track AI mentions already?

If it is Ahrefs or SE Ranking, yes, as a module, and the surfaces each one names are in the table above as read on 20 October 2026. Open the module and look for two things: which engines it covers, and whether you can put your own questions in.

Can I use my tracked keywords as prompts?

Use the ones that describe a situation and leave the head terms behind. "Project management software" is a search. "Which project management tool suits a 12 person design studio that bills hourly" is what somebody says to an assistant, and it is the one that triggers a search and names companies.

Why does the AI number in my SEO suite disagree with a reading I took by hand?

Because you asked different questions. Two honest readings of the same brand differ when the question lists differ, which is why the question count and the question file version belong next to every figure.

Is one module enough, or do I need a second subscription?

One is enough when its question list covers the way your buyers actually ask. A second one earns its place when you need your own questions, a named engine the module does not cover, or the raw answer text to show somebody who was not in the room.

What is the one thing to ask a platform before renewing?

Ask what happens to your history when you edit a question. A prompt set you can change freely without resetting the trend is worth more than a larger prompt allowance you are afraid to touch.

The question file is the part you own

A rank tracker is good at rank tracking, and the AI modules bolted onto these platforms are real products built by people who know the category. What none of them can hand you is the list of things your buyers say out loud when they are deciding. That list comes from your sales calls, and once it exists it is portable, it survives a change of supplier, and it is the only part of this measurement that gets better the longer you keep it.