NewA page for every assistant you track

Rank tracking for ChatGPT, Claude, Gemini and Perplexity

A growing share of buying decisions never touches a blue link — it ends at an assistant’s answer. FreeSERP tracks your AI visibility: it asks the questions your buyers ask, on every platform, and scores how often you are named, how often you are cited, how early in the answer you appear, and which competitors are named instead.

  • ChatGPT
  • Claude
  • Gemini
  • Perplexity
app.freeserp.com/dashboard/ai-prompt-tracker
FreeSERP AI Prompt Tracker: the brand Patagonia with multiple prompts tracked across ChatGPT, Claude, Gemini and Perplexity, an 83% average mention rate and 70% cited as a source, and a mention rate, cited rate, prominence and schedule on every prompt and platform
ChatGPT, Claude, Gemini and Perplexity
Mention rate, citations and prominence
Up to 5 answers a prompt, so it is a rate
5 credits an answer, 10 on Perplexity, 35 on Claude

Above: one brand’s prompts, scored on every platform

INTERACTIVE DEMO

Read the answers, see who gets named

An interactive example of the results screen. Click, toggle and drag — it works like the real tool.

Northpeak
northpeak.com · 3 prompts across 4 assistants · 5 answers per run
Example data
ChatGPT · answer 1 of 5 from the latest runMentionedCited

For wet-weather hiking, three shells stand out. Northpeak's Cascade 3L uses a fully taped three-layer membrane and is the most breathable on long climbs. The Ridgeline Storm Pro is the lightest at 310 g, while Summitwear's Torrent is the budget pick, though its hood adjusts less precisely.

Sources
  1. 1northpeak.comYou
  2. 2trailtestedgear.com
  3. 3ridgeline-outdoor.com
PlatformMention rateCitedProminence
ChatGPT82%60%early
Gemini68%40%late
Perplexity90%100%early
Claude74%80%mid
Who gets named instead of you
Share of voice across completed ChatGPT answers
  • NorthpeakYou34%cited in 8
  • Ridgeline29%cited in 4
  • Summitwear21%
  • Alder & Pine16%cited in 3
PlatformsThe platforms

What FreeSERP does on each assistant

Reached in whichever way gives the truest answer, and scored the same way once the answer is in.

ChatGPT

The one they ask first

The answer the product actually gave

FreeSERP reads ChatGPT's own output rather than re-creating an answer through a raw model call. What gets scored is what a person asking that question would have been shown — with web search forced on, so you are measuring AI search rather than the model's memory of its training data.

  • Real product output, not an API re-creation
  • Web search forced on, so answers are retrieved and not remembered
  • Country targeting on any prompt
  • Answers keep a link back, so you can reproduce what it said
Answer comes from
The ChatGPT product
Model
Recorded per run
Country targeting
Yes
Cost
5 credits an answer

Claude

The considered second opinion

Pinned to one model, so history means something

Claude answers through its API with web search turned on, on a version we pin deliberately. A bare model name auto-selects whatever is latest, which silently moves your numbers mid-history — this way a change in the chart is a change in your visibility, not a model that moved underneath it.

  • Pinned model version, so runs stay comparable over time
  • Web search forced on, and country targeting supported
  • Linked sources parsed out of every answer
  • 35 credits an answer — priced on what it really costs
Answer comes from
Claude API, web search on
Model
claude-haiku-4-5, pinned
Country targeting
Yes
Cost
35 credits an answer

Gemini

The one sitting next to Google

Read straight from the product, same as ChatGPT

Gemini is where a Google search increasingly ends, and FreeSERP reads its product output the same way it reads ChatGPT's. One honest limit, measured against the live API rather than assumed: Gemini rejects country targeting outright, so its answers are global — every other platform here can be pinned to a market.

  • Real product output on Gemini 2.5 Flash
  • Global answers — Gemini will not accept a country
  • Scored on exactly the same metrics as every other platform
  • 5 credits an answer
Answer comes from
The Gemini product
Model
Recorded per run
Country targeting
Not supported
Cost
5 credits an answer

Perplexity

The one that always cites

A search product, so every answer arrives with sources

Perplexity retrieves before it answers, every single time — there is no web search to switch on, because it is the whole product. That makes it the cleanest read you have on whether your pages are being used as a source rather than merely recalled, which is a different problem with a different fix.

  • Retrieves on every answer — there is nothing to force
  • Sources on every answer, resolved down to the domain
  • Country targeting on any prompt
  • 10 credits an answer
Answer comes from
Perplexity Sonar, live retrieval
Model
sonar, pinned
Country targeting
Yes
Cost
10 credits an answer
CompareSide by side

Where they actually differ

Measured against the live APIs, not assumed — including the one limitation we would rather state than paper over.

PlatformAnswer comes fromModelCountry targetingCredits an answer
ChatGPTProduct outputRecorded per runYes5 credits
ClaudeAPI, web search onclaude-haiku-4-5Yes35 credits
GeminiProduct outputRecorded per runGlobal only5 credits
PerplexitySonar, always retrievessonarYes10 credits

Web search is forced on wherever the platform allows it — Perplexity always retrieves, so there is nothing to force.

MetricsWhat you get back

What every run tells you

Brand mentions, citations, prominence and share of voice — the same metrics on every platform, so a number from ChatGPT and a number from Claude can sit in the same column.

Mention rate

How many of the run's answers named you. A rate, not a tick — these platforms answer with temperature, so a single answer flips run to run and only a rate is a measurement.

Citation rate

How often you were used as a linked source. Tracked apart from a mention and never merged into it: being named is awareness, being cited is a link, and the two take different work.

Prominence

How early in the answer your first mention lands, as a percentage of the way through. It is the closest thing AI search has to a rank position, and it works on every platform.

Share of voice

Which of the competitors you listed turned up in the same answer, and how often. Not a leaderboard you cannot verify — your rivals, counted in the answers you paid for.

Fan-out queries

The searches the assistant actually ran to build its answer. Almost nobody surfaces these, and they are the most directly actionable thing on the page: those are the terms to go and rank for, and the questions your next page should answer.

The answer itself

Every answer is stored, with your brand highlighted in the text, the sources it linked listed out, and the model that produced it recorded beside it.

PromptsWhat to track

The questions worth asking

Not keywords — questions. These are the shapes that decide a shortlist before anyone opens a search engine.

best waterproof rain jacketpatagonia vs arc'teryxalternatives to [competitor]cheapest CRM for a small teamis [brand] any goodwho should I use for [service]top 5 [category] in 2026[category] for beginnerswhat do people say about [brand]recommend a [product] under $100
SetupHow it works

A few steps, then it runs itself

01

Add your brand

Name, domain, and up to 25 aliases and 25 competitors. Terms too generic to prove anything are reported back and left out of the score, so an ordinary word your name is spelled with never counts as a mention.

02

Write the prompts

The questions your buyers actually type, up to 500 characters each and up to 250 per brand. Paste a whole list at once rather than adding them one at a time.

03

Pick platforms and samples

Any of the platforms, per prompt, and 1 to 5 answers each run — three by default. More samples is a steadier rate; fewer is a cheaper one.

04

Set a cadence

Daily, every three days, weekly or monthly — or leave it manual and run it yourself. Weekly is the default because AI answers drift slowly enough that daily mostly buys noise.

All
AI platforms, one dashboard
5
answers a prompt, every run
250
prompts per brand
$19
a month for 2,000 credits
WhyOne platform

Built to be read, not just collected

An AI visibility number is only worth having if you can see the answer behind it, turn the gap into a fix, and get it out of the tool when the report is due.

A page per platform

The brand view answers 'how is this brand doing'. The platform view answers the other one — how am I doing on Claude — across every brand you track, in one table.

Runs without you

Give a prompt a cadence — weekly is the default — and it re-runs itself. Movement lands in the dashboard, so you can see whether a change you shipped moved the numbers.

Per-market answers

Pin a prompt to a country on ChatGPT, Claude and Perplexity, and see how the answer changes between the markets you sell in.

Read the actual answer

Every stored answer is there in full, with your brand marked up in it, so you can see the sentence you were named in rather than a number claiming you were.

Export anything

CSV export on the tables, so the numbers live in your reporting stack rather than behind our login.

One balance, every tool

On any paid plan or credit pack, the same credits run Google, Maps and YouTube rank tracking, site audits and keyword research. No tool is reserved for a higher tier.

PricingCredits

Choose the plan that fits

AI visibility is on every paid plan and credit pack, with rank tracking, audits and keyword research on the same balance. Free covers Google rank tracking.

Free

Google rank tracking only

$0/mo

100 credits for Google rank checks

Get started

Features included:

  • Google rank tracking
  • Manual checks and rank history
  • 100 credits every month
  • No credit card required

Starter

One site, checked every day

$19/mo

2,000 credits — about 400 AI answers

Get started

Features included:

  • All four AI platforms
  • Mentions, citations, prominence
  • Recurring runs on a cadence
  • CSV export
  • Cancel anytime

Agency

Client work and bigger keyword sets

$99/mo

15,000 credits — about 3,000 AI answers

Get started

Features included:

  • All four AI platforms
  • Mentions, citations, prominence
  • Recurring runs on a cadence
  • CSV export
  • Cancel anytime

Not sure how many answers you need?

A run is prompts × platforms × samples. Tell us how many prompts you want to watch, on which assistants and how often, and we will size the balance for you.

Contact us
Use caseCustomers

See why teams switch

I dropped a $299/mo SEO tool the day I found FreeSERP. The free plan sold me in a week, and $19/mo covering rank tracking, audits and keyword research is a no-brainer.
Daniel CooperIndie SEO Consultant
The competitor gap report alone saved us weeks of manual research. We're now ranking on terms we didn't even know our rivals owned.
Liam ParkerGrowth Marketer
Daily rank tracking across 14 countries without touching a credit card. The CSV exports plug straight into our reporting stack.
Ethan RossSEO Operations Manager
FAQAnswers

Frequently asked questions

Anything not answered here — support@freeserp.com.

What does the AI Rank Tracker actually do?

It measures your AI visibility. It asks the AI assistants the questions your buyers ask, on a schedule, and scores what comes back. For every prompt you get the share of answers that named your brand, the share that linked to you as a source, how early in the answer you appeared, which competitors were named alongside you, and the full text of every answer it collected.

What are GEO and AEO, and why do they matter in 2026?

Generative engine optimization (GEO) and answer engine optimization (AEO) are two names for the same job: getting your brand named, and ideally linked, inside the answers AI assistants write. They matter because more and more research that used to happen across a page of search results now happens inside a single answer. If that answer recommends a handful of brands and you are not one of them, you were never on the shortlist. You cannot improve that without measuring it first, which is what this tracker is for.

How do AI assistants decide which brands to mention?

Broadly in three steps. They retrieve: the assistant searches the web and can only use pages it is able to reach and read. They weigh trust: whether what your site says about you matches what other sources say, and whether those sources look credible. Then they name: only a few brands make it into the final answer. The tracker shows you the outside of that process — the searches the assistant ran, the sources it linked and whether you made the list — so you can tell which step you are dropping out at.

Why these platforms?

ChatGPT, Claude, Gemini and Perplexity are where AI answers are actually being read, and each is reached in the way that gives the truest result: ChatGPT and Gemini are read from their own product output, while Claude and Perplexity are answered through their APIs on a pinned model version with web search on.

Does it track Google AI Overviews or AI Mode?

Not in this tracker — it covers ChatGPT, Claude, Gemini and Perplexity, and Gemini is Google's own assistant. If you want to know whether Google AI Overviews cites your site, FreeSERP's AEO & GEO audit puts real questions to AI Overviews alongside the other engines, and it runs on the same credit balance.

How do I get my brand into AI answers?

The tracker tells you where to aim; the work happens on your own site and around the web. A low mention rate on a prompt usually means no page of yours answers that question directly, so write one that does, in clear, quotable sentences. Fan-out queries show the exact searches the assistant ran, which tells you what that page needs to rank for. Cited sources show which sites the assistants trust in your category — that is where reviews, listings and mentions are worth earning. And keep the basics right: the same brand facts everywhere, structured data on key pages, and AI crawlers allowed in. The AEO & GEO audit checks crawler access and schema for you.

How will I know it's working?

Put your prompts on a weekly cadence and watch mention rate, citation rate and prominence move, platform by platform. Each run collects several answers per prompt, so a change in the chart is a change in the rate rather than one lucky answer, and Claude and Perplexity run on pinned models, so a shift there reflects your visibility rather than a model swap. Share of voice tells you whether you gained on the competitors you listed or the whole category moved.

What does it cost in credits?

Five credits per answer on ChatGPT and Gemini, ten on Perplexity and thirty-five on Claude — Claude costs us about six times as much an answer, and pricing it apart stops the cheap platforms subsidising it. A run is one prompt × the platforms you picked × your samples per run, so one prompt on every platform at three samples each is 165 credits.

Why use a tracker instead of asking ChatGPT myself?

Because these models do not answer identically twice. Ask once by hand and you learn what one roll of the dice said, on one platform, with no history behind it. The tracker asks each prompt up to five times per platform, on a schedule, stores every answer, and turns them into a mention rate you can compare with last week's and with your competitors'.

Can I track a specific country?

On ChatGPT, Claude and Perplexity, yes — set a country on the prompt and the answer is retrieved for that market. Gemini rejects country targeting at the API level, so Gemini answers are global. We would rather say that than quietly return a global answer labelled as a local one.

Is this the same as Google rank tracking?

No, and it does not replace it. There is no position 1 to 10 in an AI answer — there is being named or not, being cited or not, and how early you turn up. FreeSERP tracks both worlds on one account and one balance: Google, Maps and YouTube positions beside AI visibility.

Do I need a credit card to start?

For the AI Rank Tracker, yes: it is on paid plans, from $19/month for 2,000 credits, or with a credit pack. The free plan needs no card and covers Google rank tracking with 100 credits a month. Plans can be cancelled at any time — email support@freeserp.com if you want a hand sizing one.