Depra is an AI visibility tracker: it asks the questions your buyers type into ChatGPT, Gemini, Perplexity and Google AI Overviews, from inside your chosen country, and records whether each answer names your brand, in what position and tone, and which websites it cites. ChatGPT (with web search) and Gemini are captured from their consumer web apps. AI Overviews, the AI summary Google shows at the top of some search results, are captured from real Google results pages. Perplexity answers come from Perplexity's own answer model (Sonar) with live web search, which can differ from the answer in the Perplexity app. ChatGPT, Gemini and AI Overviews are checked daily and Perplexity weekly.
The same kind of tool is also sold as an AI rank tracker, an AEO tracker or an LLM monitoring tool. LLM stands for large language model, the kind of AI behind ChatGPT and Gemini. Whatever the label, it should tell you three things: how often AI names you, how you compare with competitors, and which websites shape those answers. Depra does all three for you and 3 to 20 competitors per project, depending on plan, and every number is computed from answers it has stored in full.
This page covers what Depra measures, which engines and countries it tracks, how competitor benchmarking and question research work, how accurate the numbers are, and what each plan costs. The full scoring rules are on how Depra measures AI visibility.
What does an AI rank tracker measure?
An AI rank tracker measures how often AI answers name your brand and where you come when they list options. Depra reads every stored answer and records these numbers for you and for each competitor you track.
- Visibility: the share of answers to your buyers' questions that name your brand, per engine, over any date range you pick.
- Average position: where you come in the list when AI recommends options. A position of 1 means you were named first.
- Sentiment: whether the answer talks about you warmly, neutrally or critically, with the exact sentence that names you so you can check it.
- Share of voice: your share of all brand mentions in the answers, next to each competitor. See AI share of voice explained for the calculation.
- Change over time: each number compared with the previous period, last month or last year.
Views that sit behind the scores
- One question at a time: for any tracked question, see its visibility trend, which engines name you, the websites and pages cited, and the answers themselves.
- AI Overview appearance: how often Google shows an AI Overview at all for your searches. When no Overview appears, Depra stores that separately and keeps it out of your visibility, so a missing Overview never counts against you.
Which AI engines and countries does Depra track?
Depra tracks four AI engines: ChatGPT, Gemini, Google AI Overviews and Perplexity. For ChatGPT and Gemini it records the answers real people see in the consumer web apps.
| Engine | What Depra records | How often |
|---|---|---|
| ChatGPT | The answer shown in the chatgpt.com web app, with web search switched on so it cites sources | Daily |
| Gemini | The answer shown in the gemini.google.com web app | Daily |
| Google AI Overviews | The AI Overview on a real Google results page for that search in that country, its source links, and whether an Overview appeared | Daily |
| Perplexity | The answer from Perplexity's own answer model (Sonar) with live web search, set to your country. It can differ from the answer in the Perplexity app | Weekly |
Asked the same 10 buying questions from inside India, Gemini cited sources in 90.0% of English answers and in 41.3% of Hinglish answers. (Depra English vs Hinglish AI shopping study, 480 answers, fieldwork 14 Aug 2026)
Countries and languages
Each question is asked from inside the country you choose, at country level. Depra covers 40 markets: India, the United States, the United Kingdom, Ireland and 29 other European countries, 6 Gulf countries and Australia. Any market is available on any plan, with up to 10 markets per project. Each plan allows a set number of tracked questions, called prompts, and a question tracked in 2 markets counts as 2 prompts.
Every plan tracks English and Hinglish questions. Hinglish is Hindi and English typed in English letters, such as 'sabse accha protein powder kaunsa hai'. Your report splits visibility between English and Hinglish, and Hinglish AI visibility tracking explains why the split matters.
What Depra does not track
- Google AI Mode, Claude, Microsoft Copilot, Grok, DeepSeek and Meta AI.
- Questions typed in Hindi script. Hinglish in English letters is tracked.
- City-level location, personas or answers shaped by a logged-in chat history. Every check is a fresh question from that country.
How do you compare your visibility with competitors?
Depra asks the same questions for you and your competitors, so every brand is scored on the same answers. A ranking table lists you and each tracked competitor by visibility, sentiment and average position, for each engine and date range.
- Suggested at signup: Depra proposes likely competitors from your website and category. You keep, remove or add to the list.
- Brands AI keeps naming that you don't track: any brand that appears in at least 2 of your stored answers is offered as a competitor you can add in one click. A suggestion you dismiss stays dismissed.
- Competitors per project: 3 on Starter, 10 on Growth, 15 on Scale and 20 on Enterprise.
- Fair start dates: a competitor is scored only on answers collected after you started tracking it.
From benchmark to next step
The ranking tells you who AI names. Citation gap analysis shows why: the websites AI cites when it recommends your competitors and never when it names you. Those gaps feed your AI action plan, a ranked list of places to get mentioned and pages to publish.
Which questions should you track for AI visibility?
Track the buying questions your customers ask AI before they choose, written without your brand name. Each one counts as a prompt in your plan, and Depra drafts a starting set when you create a project.
- 5 topics with 8 questions each: 40 suggested questions in total, mixing 'recommend' and 'compare' requests with 'which' and 'what' questions, mostly with buying intent.
- No brand names: none of the suggested questions contain your brand. Questions that already name you are left out of your scores by default, because asking about a brand by name almost always returns that name.
- Written for your market: with one market the question names it ('in India', 'in the UAE'). With several markets the place is left out and each check is asked from inside each country.
- Hinglish versions: when India is one of your markets, each suggestion also comes in Hinglish.
- Intent labels: each question is tagged Commercial, Informational or Transactional automatically.
- Full control: add, edit, pause, remove and tag questions, one at a time or in bulk. Edits keep earlier versions.
A limit to know about
Depra does not estimate how many people ask each question. Pick questions from what your sales team hears, your site search and the words buyers use in reviews, then let the tracked answers show which ones matter.
Can you read the AI answers behind each score?
Yes. Depra stores the full text of every answer it collects. For each answer you see the brands it named, in what order, the sentence that names you, and the websites it cited.
- Filter and export: filter the answers, for example by engine, brand or cited website, and export them to CSV for Excel or your own analysis.
- Cited websites, labelled: every website AI cites in your category is labelled editorial, company site, forum, reference, marketplace, directory, government or your own site.
- A report per project: one printable page with a plain-English summary, visibility by engine, English vs Hinglish, competitors and your top fixes, which you save as a PDF from your browser.
How accurate is AI visibility tracking?
AI visibility tracking is as accurate as the number of answers behind each score. AI engines can name different brands when the same question is asked twice, as covered in whether ChatGPT gives the same answer to everyone, so one check on one day can mislead.
Depra reruns the same questions on a schedule. Your report shows the number of answers behind each score and a 95% confidence range for visibility: the band the true figure most likely sits in. At 50% visibility, 20 answers give a range of about 30% to 70%. 400 answers narrow it to about 45% to 55%.
The daily brief is an in-app page that lists what changed since the previous scan day: visibility up or down, rank changes, new or lost cited websites, new brands appearing and sentiment shifts. It reports a visibility change only when the two days' 95% ranges do not overlap and each day has at least 5 answers, so ordinary run-to-run noise stays out. Depra does not send alerts by email or chat; you open the brief in the app.
Every mention also carries its quoted sentence. If a score looks wrong, open the answers and read them.
What does Depra's AI visibility tracker cost?
Plans start at ₹1,999 a month plus GST, and every plan starts with 7 days free on the plan you pick, with no card. Your first scan starts at signup.
| Plan | Price a month | Prompts | Projects | Competitors |
|---|---|---|---|---|
| Starter | ₹1,999 | 20 | 1 | 3 per project |
| Growth | ₹7,499 | 100 | 3 | 10 per project |
| Scale | ₹12,499 | 180, pooled | 25 | 15 per project |
| Enterprise | ₹25,000, invoiced | 400, pooled | 50 | 20 per project |
What a plan buys in answers
On Starter, 20 questions in one market come to about 1,900 stored AI answers a month across the four engines when every scheduled check runs. Each plan has a monthly ceiling on checks: 2,400 on Starter, 11,000 on Growth, 19,000 on Scale and 40,000 on Enterprise. If you reach it, scheduled checks drop to weekly for the rest of the month, so data keeps coming in. A one-time pack of 1,500 extra checks costs ₹1,499.
Self-serve plans are paid by card or UPI AutoPay, with a GST invoice that carries your GSTIN. Enterprise is invoiced.