Docs
Monitoring

Platforms

The AI platforms Heading tracks and how each is monitored

Heading runs your active prompts against each tracked platform once per UTC day and analyses every response for mentions, citations, position, and sentiment.

Available platforms

PlatformWhat it isCitation behaviour
ChatGPTOpenAI's conversational AI: the largest consumer AI platformSource links on web-informed answers
PerplexityAI-native search engineThe most citation-heavy platform; nearly every claim carries a source link
Google AI OverviewGoogle's AI-generated summaries above search resultsCites the indexed pages that informed the summary
Google AI ModeGoogle's conversational search experience, separate from the AI Overview above the resultsCites the pages behind the answer. Answers in English in every location
GeminiGoogle's assistant, answering outside searchLists the sources behind the answer
Microsoft CopilotThe AI summary Bing returns above its resultsCites the pages the summary drew on

Choosing your three

A property tracks any three of the six. The choice is the same on every plan, including a free snapshot property: plans differ by prompt quota, not by platform access, and no platform costs more to track than another.

  • Where the selection lives: Settings → AI models, on the property
  • What it should reflect: the platforms the property's buyers use, not a ranking of the six
  • Example selections: a local services brand on ChatGPT, Google AI Overview and Copilot; a developer-tools brand on ChatGPT, Perplexity and Gemini. Both cost the same

Swapping a platform changes what the Overall line averages, so a swap is recorded and marked on your visibility chart. Per-platform history is never deleted or hidden: deselecting a platform stops new collection, and everything already collected stays.

Update cadence

  • Paid properties: every active prompt × every tracked platform, once per UTC day. Missed fetches are detected and re-run automatically.
  • Free properties: a point-in-time snapshot across the property's tracked platforms, measured once. See Your free snapshot report.

Not tracked

Claude and GPT 5 Search appear in the AI search landscape but are not available for daily tracking. Each of the six above is read through an endpoint costing a fraction of a cent per response. Claude has no such endpoint, so reading it costs more per response than the six do, and we would rather leave it out than price every property for it. GPT 5 Search has no read path at all. Docs will be updated if either changes.

Platform coverage in your data is honest: a metric a platform cannot support is shown as not available, never approximated.

When a platform gives no answer

Some platforms answer some questions without any AI summary at all: Google shows no AI Overview, Bing shows no Copilot answer. Heading records that as a day the platform gave no answer, which still counts as a day your brand was not surfaced. A prompt that rarely draws an AI answer is a prompt worth replacing.