AI Visibility Tracker: Brand Mentions in Perplexity & Gemini
Pricing
from $20.00 / 1,000 answers
AI Visibility Tracker: Brand Mentions in Perplexity & Gemini
Ask Perplexity and Google Gemini models the questions your buyers ask, with live web search on. See if your brand is named, before or after which competitors, and which websites each answer cites.
Pricing
from $20.00 / 1,000 answers
Rating
0.0
(0)
Developer
Quiet Signals
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
2 days ago
Last modified
Categories
Features
Share
People now ask AI assistants which company to hire or which product to buy. This tool asks those questions for you and tells you whether your brand is in the answers.
You give it a brand, a list of questions and, if you like, a list of competitors. It sends each question to Perplexity's and Google's AI models with live web search switched on, and returns one row per answer:
- whether the answer names your brand, and how many times
- whether your brand is named before or after each competitor
- which competitors the answer names
- which websites the answer cites as sources, and whether yours is one of them
- the full text of the answer, so you can check every figure yourself
A summary gives the share of answers that name your brand, cite your website and name each competitor, overall and per engine.
How the answers are obtained
The questions go to each company's own model through its official programming interface, with that company's own web search: Perplexity Sonar, and Google Gemini with Google Search. Nothing is scraped and no website is automated.
This matters for how you read the results. These are the answers the models give through their interfaces, not screenshots of the Perplexity or Gemini apps. The apps add their own settings, your location and your history, so an answer you get by typing the same question into an app can differ.
Each question is sent exactly as you write it, with no instructions added, so the model answers it the way it would answer anyone.
How the counting works
No AI is used to score the answers. A brand counts as named when its name appears in the answer as a whole word, in any letter case, with spaces and hyphens treated alike ("Roto-Rooter" also finds "Roto Rooter" and "RotoRooter"). Add other spellings in the input if your brand has them.
That makes every number checkable against the answer text. It also means the tool does not judge tone: an answer that warns people away from a brand still counts as naming it. Read the answers that matter.
Input
| Field | What to enter |
|---|---|
| Brand name | The brand to look for |
| Questions | One per line, worded the way a buyer would ask |
| Competitors | Optional. Other brands to look for in the same answers |
| Other spellings of the brand | Optional. Short forms or former names that should count |
| Brand website | Optional, for example example.com. Used to tell whether an answer cites your site |
| Engines | Perplexity and Google Gemini by default |
| Answers per question | 1 to 5. See "Answers change" below |
Output
One row per answer:
| Field | Meaning |
|---|---|
prompt | The question |
engine, model | Who answered: perplexity or google, and the exact model |
sample | Which asking of the question this is, when you ask more than once |
brandMentioned | True if the answer names the brand |
brandMentions | How many times |
brandPosition | Place among the brands you track, in order of first mention. 1 means your brand was named before every competitor. Empty if it was not named |
competitorsMentioned | Competitors named, in order of first mention |
brandCited | True if the answer cites a page on your website |
citedUrls, citedDomains | The sources the answer cites |
answer | The full answer |
checkedAt | When the run started, UTC |
The run's key-value store holds a SUMMARY record with the shares per engine.
Example row, shortened:
{"prompt": "Who are the best companies for emergency drain cleaning in the US?","engine": "perplexity","model": "perplexity/sonar","brandMentioned": true,"brandMentions": 3,"brandPosition": 1,"competitorsMentioned": ["Mr. Rooter"],"brandCited": true,"citedDomains": ["dirjournal.com", "rooterhero.com", "modernize.com", "rotorooter.com"],"answer": "The best companies for emergency drain cleaning in the US are typically the national brands..."}
What it costs
There are two charges, and both appear on your Apify invoice.
- This tool's price per answer, shown on this page. Answers that fail are not charged.
- Model usage, billed by Apify's built-in model service at the model maker's price. In our test runs on November 16, 2026, a Perplexity answer cost about $0.006 and a Gemini answer $0.04 to $0.05 (Gemini runs several web searches per answer, and Google charges per search). Six answers, three from each, cost $0.16 in model usage. Your figures will vary with the question.
On Apify's free plan, model usage is billed at ten times that rate. That is Apify's rule for its model service, not ours: those six answers used $1.60 of free credit in our test. On the free plan, start with Perplexity only and one or two questions.
You need no account or key with Perplexity or Google.
Answers change
Ask an AI model the same question twice and you can get two different answers, with different brands in them. One answer tells you what the model said once. To learn how often your brand is named, set "Answers per question" to 3 or 5 and read the share in the summary. Each extra answer is charged like any other.
To follow a brand over time, run the tool on an Apify Schedule with the same questions and compare the summaries.
Limits
- OpenAI is not available today. The engine is in the list, but as of November 16, 2026 OpenAI refuses requests that come through Apify's model service. If you select it while that lasts, it is skipped, nothing is charged for it, and the other engines still run.
- Two engines by default, not the consumer apps, as explained above.
- Gemini's sources come as Google redirect links.
citedDomainsshows the real websites;citedUrlsfor Gemini holds the redirect links, which lead to the real pages. - A model sometimes answers without searching the web. That row then has no sources. It happened in one of six answers in our test.
- Name matching is literal. A brand whose name is a common word ("Apple", "Target") will be counted when the word is used in its ordinary sense.
- No tone or sentiment scoring, no country setting, and no comparison with earlier runs yet.
- Answers are limited to about 1,500 tokens (roughly 1,000 words). A longer answer is cut off at that point.
FAQ
Is this the same as what I see in the Perplexity or Gemini app? Not exactly. It is the same company's model with the same company's web search, asked through its interface. The apps can answer differently.
Why does my brand show up in one run and not the next? Because the models do not give the same answer every time. Use several answers per question and look at the share.
Can I check a result by hand? Yes. Every row carries the full answer. Switch on "Save raw responses" to keep each model's complete response in the run's key-value store as well.
Something is wrong or missing. Open an issue on this page's Issues tab.