---
title: "How to Track AI Visibility and What It Costs"
seoTitle: "How to Track AI Visibility and What It Costs"
description: "Track AI visibility with free Google and Bing reports, a manual prompt panel, and paid trackers from $29 a month. See budgets for 20- and 150-person firms."
datePublished: "2026-10-03T17:53:00Z"
dateModified: "2026-11-01T19:34:00Z"
category: "ai"
imageAlt: "Marketing lead comparing a printed cost spreadsheet with a laptop dashboard at a standing desk at night"
tags: ["ai-visibility", "ai-search", "measurement", "tool-pricing", "chatgpt"]
faq: true
headQuery: "how to track ai visibility"
featuredImageFile: "/blog/art/en/measure-ai-visibility-hero.1fe233b5.webp"
---
AI visibility tracking has three cost lines: free first-party reports, a manual prompt panel paid for in staff hours, and paid trackers starting at $29 a month and reaching custom enterprise quotes. To track AI visibility, start with Google's Search Console Generative AI report, Bing Webmaster Tools AI Performance, and the GA4 AI Assistant channel, all available at no charge. A manual panel samples answers yourself; Otterly Lite starts at $29 a month, while Profound publishes no enterprise price. Prices here are in USD and were read on 1 November 2026.
A one-off brand check captures a moment. Ongoing tracking repeats a defined set of buyer questions over time, then counts how often answers name or cite the company. The distinction matters because an answer can change when you ask the same question again.
## Can you track AI visibility for free?
Free AI visibility reports count Google impressions, Bing citations, and visits referred by AI assistants. Each answers a different question, and none is a complete count of brand mentions across assistants.
**Search Console Generative AI report: $0.** Google counts impressions of links to your site in AI Overviews and AI Mode, with breakdowns by page, country, device, and date. The report rolled out to all sites worldwide on 31 August 2026, although Google says it may not appear for a site without enough generative AI impressions. My reading of [Google's report documentation](https://support.google.com/webmasters/answer/16984139?hl=en) is that this report lists neither clicks nor queries.
When Search Console numbers move, use the [guide to measuring search performance and diagnosing decay](/guides/seo/measuring-seo-and-fixing-decay) to distinguish a meaningful change from noise.
**Bing Webmaster Tools AI Performance: $0.** The free tool reports citations, average cited pages, and a sample of grounding queries used to retrieve cited content across Microsoft Copilot and Bing AI summaries. The report has been in public preview since 10 February 2026, and [Microsoft's announcement](https://blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview) lists no click metric.
**GA4 AI Assistant channel: $0.** The channel counts referral sessions from sources such as ChatGPT, Gemini, Deepseek, Copilot, and Grok. Clicks from Google's AI Overviews and AI Mode remain in Organic Search, according to [Google's channel definitions](https://support.google.com/analytics/answer/9756891?hl=en). For context on what published studies report about those visits, see [AI word of mouth and referral traffic](/blog/ai-word-of-mouth).
A referral session tells you someone arrived at your site; a brand mention tells you an answer named the company. As far as I can tell, OpenAI, Anthropic, and Perplexity do not offer site owners a report showing how often ChatGPT, Claude, or Perplexity name them in answers. Manual panels and paid trackers sample that missing view.
## How much do AI visibility tools cost each month?
Paid AI visibility tools range from $29 a month for a small prompt panel to a custom enterprise quote. The table uses vendor prices read on 1 November 2026, with monthly billing in USD. Prompt allowances, engine coverage, and the unit being purchased differ, so the lowest price is not necessarily the lowest price for your panel.
| Tracker | Monthly price and allowance | Coverage and conditions |
|---|---|---|
| Otterly | Lite $29 for 15 prompts; Standard $189 for 100; Premium $489 for 400 | Daily tracking on four included engines: ChatGPT, Google AI Overviews, Perplexity, and Microsoft Copilot. AI Mode or Gemini costs $9, $59, or $149 each on the respective tiers; Claude costs $29, $109, or $439. |
| [Peec](https://peec.ai/pricing) | Starter $95 for 50 prompts; Pro $245 for 150; Advanced $495 for 350 | Daily tracking on three models chosen from ChatGPT, AI Mode, AI Overviews, Microsoft Copilot, Gemini, and Naver AI. Extra models, including Perplexity, are sold as add-ons; no add-on price is shown on the static pricing page. Claude is listed only on Enterprise. |
| [Semrush AI Visibility Toolkit](https://www.semrush.com/kb/1493-ai-visibility-toolkit) | $99 for 25 prompts; add 50 for $60 | Daily Prompt Tracking covers platforms including ChatGPT and Google AI Mode. No free trial. |
| [Ahrefs](https://ahrefs.com/pricing?currency=usd) | Prompt packages: $50 for 2,500 checks, $100 for 7,000, or $250 for 25,000 a month, added on top of a plan's quota. Lite costs $129 a month and includes 150 checks on AI prompts. | One check is one prompt on one platform in one location. Brand Radar AI starts at $199 and combines a research database of more than 525 million prompts with tracking for your own prompts. |
| [Profound](https://www.tryprofound.com/pricing) | Free seven-day trial; enterprise pricing by demo | The trial runs 50 prompts daily on ChatGPT, Gemini, and Google AI Overviews. Profound publishes no enterprise price. |
Otterly and Peec advertise a 15% annual-billing discount; Ahrefs advertises savings of up to 17%. Those discounts change the annual bill, so compare billing terms before treating a monthly equivalent as a monthly commitment. If you need to decide whether measurement deserves a separate budget line, the [GEO and SEO budget comparison](/blog/geo-vs-seo) addresses that allocation.
## Convert prompt limits into answers per month
Answers per month provide a common unit for comparing daily prompt allowances. One answer is one run of one prompt on one engine, so a tracker with more included engines can produce more answers from the same number of prompts.
:::formula{label="Answers sampled per month, our arithmetic"}
answers_per_month = prompts x engines x runs_per_month
:::
:::figure{src="/blog/art/en/measure-ai-visibility-fig.677b7377.webp" alt="Monthly answer samples from free reports, manual checks, and paid tracking, alongside their costs" width="1440" height="810"}
Free reports supply first-party counts, while manual panels and paid trackers sample answers. Compare the number of samples with the hours or subscription needed to produce them.
:::
Using a 30-day month, my arithmetic puts Peec Starter at 50 prompts x 3 models x 30 runs, or 4,500 answers for $95, about 2.1 cents per answer. Peec also states the 4,500-answer figure on its pricing page. Otterly Lite yields 15 prompts x 4 included engines x 30 runs, or 1,800 answers for $29, about 1.6 cents per answer.
Those unit costs assume you use every included prompt and engine. They exclude staff time to choose prompts and read results, and they do not make two different engine sets interchangeable. A ChatGPT and AI Mode panel cannot tell you what Perplexity or Copilot answered.
## Runs needed before a visibility percentage means anything
A single AI answer is an observation, not a reliable estimate of how often an assistant will name a brand. A visibility percentage divides answers that name the company by answers tested, for a defined set of prompts, engines, and dates.
In a [SparkToro study published on 27 January 2026](https://sparktoro.com/blog/new-research-ais-are-highly-inconsistent-when-recommending-brands-or-products-marketers-should-take-care-when-tracking-ai-visibility/), 600 volunteers ran 12 prompts 2,961 times across ChatGPT, Claude, and Google's AI during November and December 2025. The study was not peer reviewed. It found less than a 1 in 100 chance that any two of 100 responses from ChatGPT or Google's AI would give the same brand list; Rand Fishkin's takeaway was to ask at least 60 to 100 times and average the results.
My arithmetic: daily tracking produces about 30 runs per prompt per engine in a 30-day month. One prompt on one engine therefore reaches 60 to 100 runs in roughly two to four months; pooling prompts produces a larger sample sooner, but the result then describes that group of prompts. Weekly checks yield about four runs per month, so a monthly percentage for one prompt and engine rests on very little evidence.
## A 20-person company: hours versus a small subscription
A hypothetical 20-person regional services company tracks 10 buyer prompts across ChatGPT, Gemini, Perplexity, and AI Mode. Assume the free reports take one hour to set up once and cost $0 in subscription fees.
For a manual panel, assume three runs per prompt per engine each month, one minute to run and record each answer, and an internal staff cost of $60 an hour. My arithmetic is 10 x 4 x 3 = 120 answers, or two hours: $120 a month and $1,440 a year. Three runs give a count from 0/3 to 3/3 for each prompt and engine; that is too thin to treat as a useful percentage.
For a first pass that writes and tests the questions, use the [one-off AI brand check](/blog/does-ai-know-you). My opinion is to run the manual panel for the first month, then buy the smallest tier that covers the settled questions: you need to know what you are measuring before paying to repeat it.
Otterly Lite costs $29, with AI Mode and Gemini adding $9 each. My arithmetic puts the subscription at $47 a month or $564 a year, with about 30 runs per prompt per engine in a 30-day month. Lite also includes AI Overviews and Copilot; assume another hour a month to read the results, costing $60 at the same internal rate.
## A 150-person company: compare coverage and analyst time
A hypothetical 150-person B2B company tracks 60 buyer prompts in one country. The free Google, Bing, and GA4 reports still cost $0 in subscription fees; the following paid options cover different numbers of engines, so their totals are not equivalent coverage.
| Option | Monthly cost | Yearly cost, our arithmetic |
|---|---:|---:|
| Semrush: 25 prompts for $99 plus 50 for $60; ChatGPT and AI Mode | $159 | $1,908 |
| Ahrefs: Lite at $129 plus Growth prompt package at $100 | $229 | $2,748 |
| Peec Pro: 150 prompts on three chosen models | $245 | $2,940 on monthly billing; about $2,499 with the 15% annual discount |
| Otterly Standard: $189 plus AI Mode and Gemini at $59 each | $307 | $3,684 |
For Ahrefs, my arithmetic is 60 prompts x 2 platforms x 30 days = 3,600 checks. That exceeds the Basic package's 2,500 checks, while the Growth package's 7,000 covers the panel; this calculation leaves aside Lite's 150 included checks.
Assume an analyst spends three hours a month maintaining the prompts and reading results, at an internal cost of $75 an hour. My arithmetic adds $225 a month, or $2,700 a year, roughly the size of the subscription line. That labor belongs in the budget even when the tracker runs unattended.
At the same assumed minute per answer used in the smaller scenario, my arithmetic for a manual equivalent is 60 prompts x 4 engines x 30 runs = 7,200 answers, or 120 hours a month. That is where a tracker earns its fee: it buys repeated sampling that would consume a substantial share of a person's working month.
A hypothetical company selling in three countries with about 300 prompts reaches Peec Advanced at $495 or Otterly Premium at $489 a month; my arithmetic puts Otterly at $787 with AI Mode and Gemini added. Profound requires a demo for an enterprise quote. Check whether each plan covers the specific countries, engines, and prompt allocations you need before comparing those totals.
## Where should you stop paying for more tracking?
The cost of AI visibility monitoring stops buying useful information when added coverage does not change the questions you can answer or the decisions you make. These are my rules for choosing the next tier:
- **Pay for engines your buyers use.** An extra engine earns its fee when you need its separate results, not because it appears on a price list.
- **Use daily sampling for a monthly report.** My judgment is that paying for more frequent runs adds nothing the report can use when the report is read monthly. Daily runs already accumulate toward the study's suggested 60 to 100 over a quarter.
- **Spend prompt slots on different buyer phrasing.** In the same SparkToro post, 142 volunteer responses supplied prompts for the same need; barely two looked alike, and average semantic similarity was 0.081. Answers to those prompts still drew on a relatively consistent set of brands. My judgment is that a tracked prompt list is a sample someone chose, so include how buyers describe their problem as well as the category terms they know. Phrasing variety guards against a list that only reflects your own vocabulary; it does not mean each new phrasing changes the results.
- **Keep measurements separate.** Record mentions and citations in separate columns. Keep Google's impression count, Bing's citation count, and the tracker's sampled visibility percentage as three separate numbers.
- **Treat the tracker as a sample.** Google says [no third-party tool can access its internal ranking or AI systems](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide).
- **Drop to a smaller tier when the report stops changing decisions.** If percentages move by less than the noise between runs for two consecutive quarters and nobody changed a decision because of the report, reduce the panel or subscription.
For a company that wants a prompt set written from buyer phrasing and read monthly alongside the free Google and Bing reports, [Iron Goo's AI visibility service](/services/aio) can take on that work.