Skip to content
Platform / Measure

Perplexity, Gemini and Claude tracking

Axiom GEO runs your prompts through Perplexity, Gemini and Claude and records how each one talks about your brand. Perplexity searches the live web and cites pages. Gemini and Claude are asked without web search, so their answers show what the model itself believes about you, which is a different and equally useful view.

credits per Perplexity query
20
credits per Perplexity query
credits per Gemini query
24
credits per Gemini query
credits per Claude query
32
credits per Claude query

Key points

  • Perplexity searches the live web and cites the pages it used.
  • Gemini and Claude are asked through their APIs without web search, so they answer from what the model already knows.
  • Gemini and Claude do not name sources, so a source finder helps trace where a wrong detail came from.
  • Every answer is read for mention, position, sentiment, competitors, recommendation and, where shown, cited pages.
  • Published credit rates are 20 for Perplexity, 24 for Gemini and 32 for Claude per query.

Why track more than one AI engine?

Because each engine can give a different answer to the same question. A brand can do well in one and be missing from another. If you only watch one, you only see part of the picture.

The engines also work in different ways. Some look things up on the web before they answer. Others answer from what they already know. Knowing which is which helps you read the results properly and decide what to do about them.

It also explains a common puzzle. People often ask why their brand is named in one engine and missing in another for the same question. Sometimes it is because one engine found a list that includes you and the other didn't. Sometimes it is because one engine answered from memory. The side-by-side view makes that much easier to spot.

How do Perplexity, Gemini and Claude differ?

Perplexity searches the live web and cites pages, while Gemini and Claude are asked here without web search. The table sets out what that means in practice.

EngineHow it answers hereSources shownCredits per query
Perplexity (Sonar)Searches the live web, then writes an answerYes, cited pages20
Google GeminiFrom what the model already knows, no web searchNo24
Anthropic ClaudeFrom what the model already knows, no web searchNo32

Perplexity

Perplexity looks up pages on the web and writes its answer from them. It then lists the pages it used. That makes it a close cousin of ChatGPT tracking, since both show you which sources shaped the answer. The two often choose different sources, though, so it helps to track both.

Gemini and Claude

Gemini and Claude are asked through their APIs with web search turned off. So they answer from what the model learned in training. That shows how the model itself sees your brand, with nothing looked up on the day.

This is useful in its own way. If the model describes you well, that suggests your business is widely and clearly described. If it gets you wrong or leaves you out, that points to a gap in how you are described across the web. Changes here tend to show up when the model is updated, not overnight.

How does tracking work?

It runs the same prompts through each engine you choose and reads every answer the same way.

  1. Use your prompt set. The same buyer questions run across every engine, so results are easy to compare.
  2. Pick the engines. Turn on Perplexity, Gemini and Claude alongside ChatGPT and Google AI Overviews.
  3. Run on a schedule. Each run adds to the history, which smooths out the natural variation between answers.
  4. Read each answer. Every answer is checked for mention, position, sentiment, competitors named and whether you were recommended.
  5. Keep the sources. For Perplexity, the cited pages are stored with each answer.

Answers are collected through the APIs, not by scraping the consumer apps. The apps can differ a little because of personal history, location and model version.

What do you see on screen?

You see each engine side by side, so gaps stand out. A chart shows share of voice and prominence per engine over time. You can open any answer to read it in full.

For Perplexity, the cited pages feed into AI citation tracking. For Gemini and Claude there are no cited pages, so the focus is on what they say and how often they name you.

What if Gemini or Claude get your details wrong?

That is a common problem, and it is one that AI answer accuracy checks are built for. Each answer is broken into claims and checked against your approved fact sheet. Because Gemini and Claude name no source, a source finder searches for pages that state the wrong detail. It checks the pages other engines cited for the same question, your own website and Google, and shows the matching text.

Our guide on how AI engines choose sources explains why engines with and without web search behave so differently.

Who uses it?

Brands that sell nationally tend to track all the engines, because buyers use different ones. B2B and software firms often care most about how the models describe them, since buyers ask for comparisons. Agencies use the side-by-side view to show clients where they are strong and where they are missing. Each query has a published credit price, shown on the pricing page.

Where it fits

These engines sit inside AI visibility tracking with ChatGPT and Google AI Overviews. ChatGPT visibility tracking covers the other search-enabled engine. AI answer accuracy checks catch wrong details, and the entity audit checks how clearly your site describes your business.

Frequently asked questions

How is Perplexity different from ChatGPT for tracking?

Both search the web and cite pages, so both show you which sources shaped the answer. They often choose different sources and name different brands for the same question. Tracking both shows whether a gap is specific to one engine or common to both.

Why are Gemini and Claude asked without web search?

Asking without web search shows what the model itself believes about your brand. That comes from what it learned in training, not from pages it found today. It is a useful check on how well your business is known, and it often explains answers that look out of date.

Do Gemini and Claude show sources?

No, not in the way Axiom GEO asks them, because web search is off. They answer from what the model already knows. If they get a detail wrong, the source finder in answer accuracy searches the web and your own site for pages that state that detail.

How do I optimise my content for ChatGPT, Perplexity and Gemini?

Start with clear pages that answer real buyer questions and state your facts plainly. For Perplexity and ChatGPT, the cited pages show which sources to aim for. For Gemini and Claude, being described the same way across many trusted sites helps the model form a steady picture of you over time.

Which plan includes Perplexity, Gemini and Claude?

The Starter plan includes Gemini. The Professional plan and above include ChatGPT, Claude, Gemini and Perplexity. Each query uses credits at a published rate, and every credit spent is logged, so you can see what each engine costs you.

Will results match the Perplexity, Gemini or Claude apps?

They will be close but can differ a little. The consumer apps may use personal history, location, a different model version and, for Gemini and Claude, web search. Axiom GEO asks through the APIs so every run is asked the same way and results can be compared over time.
Reviewed
Perplexity, Gemini and Claude tracking for your brand