AI Visibility Tracker

See if ChatGPT, Gemini, and Google AI Mode mention and cite your brand: rank, share of voice, competitors, and sources.

For brands, SEO and GEO agencies, and marketers who need to know what AI assistants tell customers. Every run tells you plainly what worked and what did not: no silent empty runs.

What you can do with it

• AI share of voice: run the questions your buyers ask and see how often each AI names you vs your competitors (shareOfVoice, brandRank).
• Citation tracking: brandCited and brandCitedUrls show when an AI uses your own pages as a source, and sourceDomains shows which sites it trusts instead.
• Content gaps: find prompts where competitors are mentioned and you are not (competitorsMentioned), then target the sources the AI relies on.
• Shopping visibility: brandInProducts shows whether your products appear in the product cards Google AI Mode shows with its answer.
• Monitoring: schedule the same prompts weekly and chart mentions and rank over time.

What you get

Field	What it is	Example
prompt	The question you submitted.	best running shoes for flat feet
type	answer or error.	answer
engine	chatgpt, google_ai_mode, or gemini.	google_ai_mode
engineName	ChatGPT, Google AI Mode, or Gemini.	Google AI Mode
country	Country the question was asked from, when set.
brand	The brand you track.	Brooks
brandMentioned	True when the answer names your brand.	true
brandMentions	How many times the answer names your brand.	2
brandRank	Order of first mention among your brand and competitors that appear (1 = named first).	2
brandCited	True when one of the answer's sources is your website.	false
brandCitedUrls	Your pages the AI used as sources.	[https://www.brooksrunning.com/en_us/mens/shoes/ro…]
shareOfVoice	Your mentions as a percent of all brand and competitor mentions in the answer.	25
competitors	Each competitor: name, mentioned, mentions, rank, cited.	[{"name": "Hoka", "mentioned": true, "mentions": 2…, …]
competitorsMentioned	Names of competitors the answer mentions.	[Hoka, …]
brandInProducts	True when your brand appears in the shopping product cards shown with the answer (Google AI Mode).	false
competitorsInProducts	Competitors whose names appear in the product cards.
products	Shopping product cards shown with the answer: title and link (Google AI Mode).
productCount	Number of product cards.	0
sources	Sources the AI cited or linked: title, url, domain, publisher.	[{"title": "Best Running Shoes for Overpronation i…, …]
sourceDomains	Domains of the sources, in answer order.	[runweekly.com, …]
sourceCount	Number of sources.	0
answer	The AI's full answer text. Empty when Include full answer is off.	When you have flat feet (low or absent arches), your feet tend to rol…
answerWordCount	Words in the answer.	310
model	Model name when the engine shows it.	3.5 Flash-Lite
checkedAt	When the answer was collected (UTC, ISO 8601).	2026-10-23T13:25:19+00:00

Examples come from a real run. Every row has every field; a value the page did not show is null, never guessed.

One row per prompt and engine: 3 rows per prompt with all engines on. Engines are asked in parallel; one prompt on all three takes about a minute.

How to use it

1. Click Try for free (or Start) on this page.
2. Enter one prompt per line in the first field. The form is prefilled with a working example.
3. Optional: set Max results and the other options.
4. Click Start. Open the Output tab, or export JSON, CSV, Excel, or HTML.

Example input

{
  "prompts": [
    "best running shoes for flat feet"
  ],
  "brand": "Brooks",
  "brandDomains": [
    "brooksrunning.com"
  ],
  "competitors": [
    "Hoka",
    "ASICS",
    "Saucony"
  ]
}

Example output

One real row from a run on 23 October 2026 (long values shortened here):

{
  "prompt": "best running shoes for flat feet",
  "type": "answer",
  "engine": "google_ai_mode",
  "engineName": "Google AI Mode",
  "country": null,
  "brand": "Brooks",
  "brandMentioned": true,
  "brandMentions": 2,
  "brandRank": 2,
  "brandCited": false,
  "brandCitedUrls": [],
  "shareOfVoice": 25.0,
  "competitors": [
    {
      "name": "Hoka",
      "mentioned": true,
      "mentions": 2,
      "rank": 4,
      "cited": false
    },
    {
      "name": "ASICS",
      "mentioned": true,
      "mentions": 2,
      "rank": 1,
      "cited": false
    },
    {
      "name": "Saucony",
      "mentioned": true,
      "mentions": 2,
      "rank": 3,
      "cited": false
    }
  ],
  "competitorsMentioned": [
    "Hoka",
    "ASICS",
    "Saucony"
  ],
  "brandInProducts": false,
  "competitorsInProducts": [],
  "products": [],
  "productCount": 0,
  "sources": [],
  "sourceDomains": [],
  "sourceCount": 0,
  "answer": "When you have flat feet (low or absent arches), your feet tend to roll inward upon impact—a motion known as overpronation . The best running shoes for flat f…",
  "answerWordCount": 310,
  "model": null,
  "checkedAt": "2026-10-23T13:25:19+00:00"
}

Pricing

Pay per result: $20.00 per 1,000 rows, plus $0.001 per run start. No subscription.

Event	When	Price
Actor start	Once per run (per GB of memory; the default 256 MB counts as one)	$0.001
Result row	Each row saved to the dataset	$0.02

Example	Rows	Cost
10 answers	10	$0.001 + $0.20 = $0.201
100 answers	100	$0.001 + $2.00 = $2.001
1,000 answers	1,000	$0.001 + $20.00 = $20.001

• Retries after a bot check or a temporary error cost you nothing. You pay only for rows saved.
• An input that fails after all retries saves one error row (one row charge), so you can see what failed and why.
• Set Max total charge per run in the run options. The Actor stops cleanly at that limit and keeps every row saved so far.

What you can count on

• Same columns on every row. Missing values are null, so spreadsheets and databases never break.
• No silent failures. An input that fails gets an error row with errorCode and a plain-language errorMessage. The other inputs keep running.
• A clear run summary. The run status says how many results were saved, how many inputs failed, and why. The same numbers are saved as the RUN_SUMMARY record in the run's key-value store, for automations.
• Automatic retries. Bot checks and temporary errors are retried with a fresh request before an input counts as failed.
• Limits respected. Max results and your spending limit stop the run cleanly.

Use it from code, automations, and AI agents

API (returns the rows when the run finishes):

curl -X POST "https://api.apify.com/v2/acts/datacortex~ai-visibility-tracker/run-sync-get-dataset-items?token=YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"prompts": ["best running shoes for flat feet"], "brand": "Brooks", "brandDomains": ["brooksrunning.com"], "competitors": ["Hoka", "ASICS", "Saucony"]}'

Python (pip install apify-client):

from apify_client import ApifyClient

client = ApifyClient("YOUR_API_TOKEN")
run = client.actor("datacortex/ai-visibility-tracker").call(run_input={"prompts": ["best running shoes for flat feet"], "brand": "Brooks", "brandDomains": ["brooksrunning.com"], "competitors": ["Hoka", "ASICS", "Saucony"]})
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(row)

JavaScript (npm install apify-client):

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: 'YOUR_API_TOKEN' });
const run = await client.actor('datacortex/ai-visibility-tracker').call({"prompts": ["best running shoes for flat feet"], "brand": "Brooks", "brandDomains": ["brooksrunning.com"], "competitors": ["Hoka", "ASICS", "Saucony"]});
const { items } = await client.dataset(run.defaultDatasetId).listItems();

No-code: connect the Actor in Make, Zapier, n8n, or Google Sheets through Apify's integrations, or add a webhook to get notified when a run finishes.

AI agents (MCP): add https://mcp.apify.com?tools=datacortex/ai-visibility-tracker to Claude, Cursor, or any MCP client. The agent can call this Actor as a tool with the same input as above.

Limitations

• AI answers change from run to run and by country. Treat one run as a sample; track trends over several runs.
• Mentions are whole-word matches of the names you enter, in any case. Add common spellings (for example ASICS and Asics are the same; Hoka One One is different from Hoka) as needed.
• brandRank orders only your brand and the competitors you enter, by their first mention.
• Some Google AI Mode sources are redirect links without a readable address; they keep their publisher name and have an empty url.
• Perplexity is not offered yet.

FAQ

What should I put in Prompts? Real questions your customers ask, such as best CRM for a 5-person team or is Brand X good for beginners. Ask both category questions and questions that name you.

How is share of voice calculated? Your mentions divided by all mentions of your brand and the competitors you entered, in that answer, as a percent.

Does it count my website as a source automatically? Enter your domains in Your website domains. If you leave it empty, any source domain that contains your brand name counts.

Can I ask from another country? Yes. Set Country (for example Germany). Answers and sources often change by country.

What does an error row mean? It marks an input that failed after retries. errorCode is short (blocked, empty_result, not_available, http_404, or invalid_... for a bad ID or URL) and errorMessage says what to do. Remove error rows with a filter on type.

Can I run it on a schedule? Yes. Create a task with your input in Apify Console, then add a schedule (for example every day at 8:00).

Is it legal to collect this data? The Actor reads public pages that anyone can open without logging in. You are responsible for how you use the data, including copyright, personal data rules such as GDPR, and the site's terms. If you are not sure, ask a lawyer.

Related Actors

• Google Search Scraper: Google search results as rows.
• Google Trends Scraper: Google Trends by keyword.

Support

Found a problem or need a field? Open an issue on the Issues tab. Include the run ID and one input that shows the problem. We read every issue.