SE Ranking AI Visibility Tracker: Setup, Limits and Costs

What the tracker does, and why it matters

SE Ranking tracks how brands appear in AI answers inside its normal plans, while rivals either include far fewer prompts or sell AI tracking as a separate toolkit. The AI Results Tracker shows how often a brand is mentioned or linked in AI answers, how it ranks against competitors, and which pages and domains shape those answers. It lives inside an ordinary SE Ranking project, next to classic rank tracking. No extra login. This playbook covers what the tracker, AI competitive research and SE Visible measure, what they cost and how to set them up, as of October 12, 2026.

The tracker covers Google AI Overviews, Google AI Mode, Gemini, ChatGPT and Perplexity ^[1]. You choose engines per project at setup, alongside classic search engines. Claude is not tracked as an answer engine, although SE Ranking's MCP server works inside Claude. That is generous coverage for the price. Few rivals match it. The weak spots are a confusing add-on unit, brand matching that is literal, and answers that change from one check to the next. All three are manageable once you know about them.

What Core, Growth and the trial include

Core includes 100 prompts tracked daily across 5 AI engines, unlimited AI sources tracking and AI competitive research on 5 domains ^[2]. Growth raises that to 250 prompts and 15 domains. The trial allows 20 prompts a day and AI research on 3 domains, which is enough to see how the tabs work on a small sample of your own questions. One detail shapes everything that follows: prompt limits are shared across all projects in the account, not granted per project. An agency with ten clients on Core is splitting one pool of 100 prompts between them. Plan the split before you start importing.

The AI Search add-on and its two-way unit

What the add-on costs

The AI Search add-on extends tracking limits, removes the brand cap on AI competitive research and includes SE Visible ^[3]. Its entry package costs US$89 a month month-to-month, or US$71.20 a month billed annually. The larger packages cost US$179 and US$345 a month, or US$143.20 and US$276 a month billed annually. The annual figures assume you commit for the full year.

Prompts or checks? The pages disagree

The add-on's unit is described two ways. The help center calls the packages 200, 450 and 1,000 checks per month and defines one check as one prompt on one AI platform. The pricing page labels the same packages 200 prompts, 450 prompts and 1000 prompts. Those are very different amounts, and nothing on the public pages settles which one is right.

If the unit is a monthly check, the entry package covers 40 prompts on five engines checked once a month. If the unit is a prompt tracked daily, it covers 200 prompts on every check. For scale, Core alone tracks 100 prompts daily across five engines, which could mean up to 15,000 checks a month. Ask sales which reading applies before paying. Get it in writing.

How usage is counted, and why paused projects still cost you

The help center lists two limits for the tracker. The first is the prompt list, counted account-wide. The second is monthly checks, where one prompt on one engine is one check ^[4]. Its worked example is simple: 20 prompts tracked on 3 engines use 60 checks per refresh. So every engine you switch on multiplies your usage, and five engines cost five times what one does. When checks run out, you cannot add a prompt or a website until the limit renews.

The trap that catches agencies is cleanup. Prompts in inactive projects still count against the limit, and pausing or archiving a project does not free them. Only deleting the prompts or the project releases the slots. Delete, don't pause. Clean up old client projects before you add new prompts, or the limit message will appear much sooner than you expected, usually in the middle of onboarding someone new.

Setting up a project that measures the right thing

The setup steps

Create a project, select AI engines next to classic search engines, give each one a country, a location and a brand name, then add prompts. Keywords are optional if you only want to track AI answers. Prompts can be imported from CSV or pasted, and assigned to groups in the same step, which saves a round of tidying later when you report on them.

Get the brand name exactly right

Mentions are detected by matching the configured brand name against each answer. The help center warns that a variation such as ION versus ION Group can count as no mention at all. In the tracker, "There is no alias or synonym list to configure", so you have to enter the exact form AI engines actually use. Check the spelling twice. A wrong brand name does not throw an error; it quietly reports you as invisible.

Choosing and sourcing prompts

Suggested prompts are generated by an AI model from three inputs only: the project's domain, the brand name and the language. They do not read your tracked keywords, your rankings or your search data. Treat them as a first draft, and write your own if your brand name is ambiguous or your site does not explain your niche clearly. The help center recommends five prompt types: informational, comparative, instructional, brand-specific and transactional. For about 100 prompts it suggests 30 to 40 awareness, 40 to 50 consideration and 10 to 20 brand prompts, kept separate.

Good sources for ideas are your top non-branded keywords rewritten as questions, People Also Ask boxes, Reddit and forum threads, and competitors' paid keywords of three words or more. The Insights section also surfaces prompts where your brand already appears but is not yet tracked. Brand prompts inflate visibility, so always report them on their own rather than blended into the headline number.

Reading the three tabs

Rankings, and the organic overlap

The tracker has three tabs: Rankings, Competitors and Sources. Rankings shows top-3 presence, source presence and a mention and link presence chart. A link icon opens the cached AI answer. Green cells mark your brand, blue cells mark competitors, and a grey No AIO means the prompt triggered no AI Overview. For AI Overviews only, the tracker compares the URLs cited in the overview with Google's top 20 organic results. The value is matching URLs divided by 20, so 6 matches is 30%. A high overlap means your organic pages feed the overview; a low one means the overview draws on pages you do not rank with.

Mentions, links and sources are different numbers

Link presence, mention presence and the Sources count measure different events. An answer can name your brand without linking to you, and link to you without naming you. A URL can appear in Sources while Rankings shows a dash, because Sources records every citation while Rankings only matches your project. Explain this to clients early. Otherwise they will see the numbers as contradictory.

Sources, Competitors and Insights

Sources lists the domains and pages AI engines cite for your prompts, with coverage, mention rate, Domain Trust, backlinks and first and last seen dates. It also counts competitor-only mentions. In practice, that is an outreach list. Competitors shows, for each prompt and date, which brands an answer named and which sources it listed, with a cached copy of each answer. Use it to read how an engine describes a competitor, not only how often it names them. Insights then flags untracked keywords that trigger AI Overviews and sources that mention competitors but not you, with recommendations as concrete as engaging Reddit communities.

Why one check proves nothing

The help center puts it bluntly: "AI engines do not return a fixed answer to a fixed prompt". The same prompt can come back with different sources on different dates, so its advice is to read a trend across several checks rather than a single date. Daily noise misleads. Report weekly or monthly movement to clients, not daily swings, even when a good day is tempting.

That shapes a sensible first month. In week one, set the exact brand name, choose engines and import 50 prompts in groups. In week two, read Sources and list ten pages that cite competitors. In week three, start outreach and content fixes. In week four, compare presence across the month, not the last day. A realistic target for a small brand is not a jump in visibility but a clean baseline: the same 50 prompts checked every day on the engines its customers use, with branded prompts split out, so next month's changes read as results rather than setup noise.

Competitive research and ads in AI answers

AI competitive research

AI Search in Competitive Research analyzes any domain's presence across the five engines, with up to 5 competitors and a topic-level view. Without a brand name, only domain presence, meaning links, is available. An Overlap filter shows topics where competitors appear and you are missing, which is the fastest way to find gaps before you commit prompts to tracking them.

Paid placements are now visible

AI Search now marks paid placements in AI Mode and ChatGPT answers, as part of Competitive Research. SE Ranking's own study of 50,006 commercial prompts found ChatGPT ads on 25.94% of them, with the advertiser also cited organically in just 3.63% of placements. A second study found a paid placement on 29.45% of AI Mode queries, rising from 24.33% under US$2 CPC to 53.56% at US$10 and above. The lesson is plain. Paid and organic AI visibility move independently. Budget them separately.

SE Visible, the separate platform

How it differs from the tracker

SE Visible is a separate platform for AI brand visibility, included with the AI Search add-on or sold on its own. It tracks the same five engines in 42 countries, daily or weekly, and its FAQ says it collects real AI responses through a browser interface rather than through APIs. The counting rules differ in an important way. In SE Visible, engines do not use up your prompt allowance, and each prompt counts once per location and language. In the AI Results Tracker, each engine is a separate check. So turning on all five engines is free in SE Visible and five times the cost in the tracker.

SE Visible also fixes the tracker's literal brand matching. It suggests brand aliases, such as abbreviations and product names, which you confirm under project settings, and a confirmed alias counts everywhere: visibility, share of voice, position and sentiment. Review them first. The help center calls unreviewed aliases the most common reason a project's visibility looks too low. A project tracks one brand in one location and takes about two minutes to set up.

Scoring, branded prompts and the gap score

Each mention is weighted by its position among the brands in that answer, and answers without the brand count as zero. The vendor's example: third of five, missing, and second of four average out at 45%. Net sentiment runs from minus 100 to plus 100 and is kept separate from visibility. Prompts that name a brand often push visibility close to 100%, so a control excludes branded prompts from all statistics. It is off by default. Turn it on before showing a client their numbers.

Every prompt is labeled with intent and funnel stage: awareness, consideration or decision. SE Visible advises starting with decision prompts, because being missing there costs the most, and that is a useful order for content work too. Its Sources view lists every website cited in the collected answers, with coverage, a gap score and SEO metrics. The gap score is high for pages that are cited often and mention competitors but not you, and pages that already link to you are, in the help center's words, the easiest wins.

The rest of the AI toolkit

Several neighboring tools round out the picture. SE Ranking's classic Share of Voice is separate from AI tracking and covers Google desktop only. AI Traffic Analytics pulls sessions, engagement, bounce rate and revenue from AI chatbots out of GA4, covering ChatGPT, Gemini, Perplexity, Claude and others. That links visibility in answers to visits that actually arrive, which is the proof clients ask for.

The MCP server lets Claude, ChatGPT and other assistants query AI search data in plain language. Through the API, the AI Search endpoints return brand presence, share-of-voice leaderboards against up to 10 competitors and the prompts where a brand appears. An overview call costs 800 credits and a leaderboard 7,500, so Core's 25,000 monthly credits cover three leaderboards. Our SE Ranking API guide goes deeper.

Two practical notes before you buy. The AI Overviews tracker page lets you check 5 keywords for free on desktop. And some marketing pages lag the product: the generative engine optimization page still mentions only AI Overviews and ChatGPT, and the AI visibility tracker page speaks of 7 supported markets, while SE Visible covers 42 countries. Trust the help center. It lists all five engines. Check that your market is covered in the product you buy.

How it compares, and who should use it

Semrush's AI Visibility toolkit starts at US$99 a month, and its SEO plan has no dedicated prompt quota. Ahrefs includes 5, 10 or 20 tracked AI prompts on Lite, Standard and Advanced. SE Ranking includes 100 daily prompts in its US$129 Core plan, which is the cheapest serious entry point among the three. At list price that works out at US$1.29 per prompt a month, and against Ahrefs Lite's five prompts it is twenty times the prompt capacity before any add-on. For a team that wants to watch a whole category of questions across five engines, that difference decides the choice. For the full plan math, see our SE Ranking pricing guide.

The included tracker suits agencies adding AI visibility to existing SEO reports and in-house teams that want a first read on AI answers. Brands that need strategic AI reporting across many locations should look at SE Visible through the add-on, where engines and aliases work in their favor. Teams that only want occasional checks can use the free AI visibility checker instead.

The bottom line

SE Ranking offers the most AI tracking per dollar of the major SEO suites, and it does so inside the base plan rather than as a bolt-on. The tool rewards careful setup. Configure the brand name exactly, separate branded prompts, delete rather than pause old projects, and read trends across weeks rather than single checks. Before you buy more capacity, get the add-on's unit confirmed in writing, because the vendor's own pages describe it two different ways. For the wider verdict on the suite, read our SE Ranking review.