AI Brand Watch - AI Visibility Tracker & LLM Brand Monitoring
Pricing
from $80.00 / 1,000 brand query checks
Know the day AI stops recommending you. AI visibility tracker and AI brand monitoring across ChatGPT, Claude, Perplexity & Gemini: share of voice vs competitors, brand visibility monitor, alerts only on real change (signed webhook). GEO/AEO audit, visibilityScore 0-100. $0.08/query, no subscription
Your brand's domain, e.g. "desvela.ai". Used to detect citations to your own site (ownDomainRate), to match the domain as a mention, and to feed the own-URL component of the visibility score. Optional.
Other ways AI engines may refer to your brand (domain, alternative spellings).
[]Competitor name → domain, e.g. {"Coda":"coda.io","ClickUp":"clickup.com"}. Enables competitorCitationRate: which competitor the AI cites by URL (same domain matching as your own domain: www, subdomains). Without it the field is null.
{}{ "Coda": "coda.io", "ClickUp": "clickup.com"}The 5-30 questions that define your category, e.g. "best CRM for small business in Spain". These are asked to each AI engine. Leave empty for zero-config: the actor reuses the frozen question set of a previous run, or discovers a new one for you.
Only applies when you provide queries: also discover additional category questions and combine them (deduplicated, max 30). Without queries this is ignored — zero-config mode already discovers. Discovery is charged as one 'prompt-discovery' event ($0.04) only when it actually runs.
Which AI answer engines to check.
[ "openai", "perplexity"]AI answers vary between runs. We ask each question several times and report the mention rate (fraction of samples where the brand appeared) across samples. Example: present/absent/present → mentionRate = 0.67; a single sample would have given 0% or 100% arbitrarily. The price per query×engine covers all samples of the cell. IMPORTANT: 4 and 5 apply to Perplexity and Gemini only. ChatGPT and Claude are always measured at 3 samples, because they bill each web search separately and charge the page content they read as input tokens, which makes them roughly four times more expensive per sample at the same cell price. Every row reports the samples it actually used, and the run summary lists them per engine.
POST endpoint that receives a signed event when your brand visibility changes. Leave empty to only use the dataset.
If enabled, the webhook fires only when something changed vs the previous run.
Generate 3-5 concrete GEO actions from this run's data (which sources each engine cites, where competitors win). Charged as one 'recommendation' event ($0.04) only when delivered.
Label each measured cell as positive/neutral/negative toward your brand (one Claude Haiku call per cell, fed with the verbatim evidence). Charged as one 'sentiment' event ($0.04) per delivered cell — e.g. 20 cells add $0.80. Default off: existing runs cost exactly the same. Descriptor only, never triggers change alerts.
How to treat the frozen question set for this brand. 'keep' (default) reuses the previous run's questions and compares against them — the watch mode. 'renew' deliberately starts a fresh baseline with the questions you provide now; that run establishes a new baseline and produces no change alerts.