Gemini Brand Visibility Tracker — Share of Voice & Citations

What does it do?

Ask Google Gemini the category questions your buyers ask, and see which brands it recommends. Each question ("What is the best issue tracker for a startup engineering team?") is sent to Gemini with Google Search grounding. For every answer you get whether your brand is named, its mention order against your competitors, which competitors were named, the source URLs Gemini cited and a short answer excerpt. SUMMARY adds up your visibility score and how many answers named each competitor: your share of voice in Gemini. Successful answers cost $0.12 each; failed answers are free.

This is a focused version of AI Answer Visibility Tracker: the same engine and price, set up for brand-vs-competitor comparison on category questions written without your brand name.

How to use

1. Enter Your brand, its website and up to 10 Competitors to compare.
2. Write up to 10 Category questions a buyer would ask, without your brand name. The prefilled example (Linear vs Jira, Asana and ClickUp) shows the format.
3. Run it on a paid Apify plan. Sort the dataset by brandMentioned, read competitorMentions to see who Gemini names instead of you, and open citations to see which pages it trusts.
4. Where your brand is missing, geoAuditUrl opens a GEO readiness audit for a fix plan. Save the input as a task and schedule it weekly to track SUMMARY.visibilityScore over time.

Input

• brand and domain (required): your exact brand name and public website. The domain also counts as a mention when the answer text names it.
• competitors: up to 10 exact brand names.
• queries: up to 10 category questions. Questions without your brand name measure whether Gemini recommends you. Left empty, 10 questions that name your brand are generated ("What is Linear?", "What are alternatives to Linear?", ...); those show how Gemini describes you, not whether it recommends you.
• engines: keep the default gemini.

Example:

{
  "brand": "Linear",
  "domain": "linear.app",
  "competitors": ["Jira", "Asana", "ClickUp"],
  "queries": ["What is the best issue tracker for a startup engineering team?"]
}

Output

Each answer is one ANSWER row with brand, domain, query, engine, checkedAt, success, brandMentioned, mentionPosition, competitorMentions, citations, answerSummary, grounded, geoAuditUrl and charged. Failed rows have success: false, errorType and error.

{
  "type": "ANSWER", "brand": "Linear", "domain": "linear.app",
  "query": "What is the best issue tracker for a startup engineering team?", "engine": "gemini",
  "success": true, "brandMentioned": true, "mentionPosition": 1, "competitorMentions": ["Jira"],
  "citations": ["https://..."], "answerSummary": "For a startup engineering team, Linear is ...",
  "grounded": true, "geoAuditUrl": null, "charged": true
}

The row above is an illustration (shortened excerpt and citations). In the engine's paid test on 2026-10-23 this question was answered with Google Search grounding and 9 cited sources, and the answer named both Linear and Jira.

• mentionPosition is the order of first appearances among your brand (or its domain) and the competitors in the answer text; it is not a search result rank.
• citations are the source URLs Gemini used (up to 20). Check grounded and citations before treating an answer as web evidence.
• SUMMARY (key-value store and the last dataset row) has status, queriesAttempted, successfulQueries, failedQueries, brandMentions, visibilityScore (brandMentions / successfulQueries × 100), competitors (each with its number of answers that named it), chargedEvents, notProcessed and estimatedGrossUsd.

Control your cost

$0.12 per successful Gemini answer (engine-query, the same event and price as AI Answer Visibility Tracker). A question that gets no usable answer, fails or is skipped is not charged. Answers are saved before charging; charged in the dataset means charge eligible and SUMMARY.chargedEvents is the confirmed count. A spending limit stops further questions. The most one run can cost is 10 questions × $0.12 = $1.20; the prefilled example has 3 questions ($0.36).

Live Gemini answers need a paid Apify plan: each one is a paid, search-grounded model call. On the free plan, and when the untouched example input is run as is (Apify's automated test of the form), the run returns up to three free paid_plan_required rows and makes no Gemini call. Change the brand, a competitor or a question to run live answers.

Limitations

• Gemini only (grounded with Google Search). ChatGPT, Perplexity and Claude are not covered; for them, use a tracker built for that engine.
• AI answers change between runs and between users; one run is a sample, not a ranking. Repeat the same questions over time to see a trend.
• Mentions are exact-name matches (case-insensitive, whole words); nicknames or misspellings are not counted.
• Up to 10 questions and 10 competitors per run.

FAQ

How is this different from AI Answer Visibility Tracker? Same engine and price. This version is set up for share of voice: category questions without your brand, a competitor list, and a view that puts your brand next to the competitors Gemini named.

Why did my run return paid_plan_required rows on a paid plan? The input was the untouched example, which is kept free for Apify's automated tests. Change any field and run again.

Am I charged when Gemini fails? No. Only saved, usable answers are charged.