Chat Features
Every AI Answer Has Structure. Now You Can Measure It.
Product cards, comparison tables, citations, follow-up prompts. Chat Features tracks the components AI platforms assemble into an answer, and how much of that answer is yours.
AI answers are assembled from components
Ask an AI platform a commercial question today and you rarely get prose. You get a product carousel, a comparison table, a set of cited sources, and a row of follow-up prompts. Each of those is a distinct component the platform chose to render, in a slot it chose to render it in.
The industry has spent two years measuring whether a brand gets mentioned. That was the right first question. It is no longer sufficient, because a brand named in the fourth paragraph and a brand occupying the first product card are both mentions, and they perform nothing alike.
Google search teams solved this problem years ago with SERP feature tracking. Chat Features applies the same discipline to AI answers, and as far as we are aware it is the first feature-level analytics for AI answer surfaces available anywhere.
Feature types
Product cards, comparison tables, citations, maps, image grids, reasoning disclosures, sponsored slots, and more. The taxonomy expands as platforms ship new components.
Position within the answer
Which component you appear in, and where that component sits in the rendered answer. The equivalent of Visual Rank for a chat surface.
Answer Share
How much of the rendered answer is attributable to you, weighted by the features involved rather than by word count. Share of Voice for AI answers.
Not every platform renders the same answer
The same question produces a shopping carousel on one platform, a citation list on another, and plain prose on a third. Feature support is uneven and it changes without notice, so the tracker records which platform rendered which component on every capture.
Structured components are crowding out prose
Prose is where an unlinked brand mention lives. Structured components are where clicks, citations, and purchase intent live. Watching that ratio move is the AI answer version of watching organic results get pushed below an AI Overview.
What we record for each answer
Component type and order
Which features rendered, in which slot, and in what sequence down the answer.
Occupied height
How much of the rendered answer each component takes up, so displacement is measurable rather than inferred.
Entities inside each component
Which brands, domains, and products occupy each card, table row, and citation, including your competitors.
Change against yesterday
Features that appeared, disappeared, or reordered since the previous capture, with alerting on the ones you care about.
Every chat feature we track
Grouped by the job the component does inside the answer. Support varies by platform and the tracker records which rendered where.
- Inline citations
- Source cards
- Link previews
- Domain attribution clusters
- Product cards
- Shopping carousels
- Price comparison blocks
- Merchant listings
- Agentic buy actions
- Comparison tables
- Step and how-to lists
- Pros and cons blocks
- Code blocks
- Image grids
- Video embeds
- Charts and diagrams
- Knowledge cards
- Maps and local packs
- Business profile cards
- Follow-up prompt chips
- Clarifying questions
- Reasoning disclosures
- Freshness stamps
- Sponsored slots
One dataset with your SERP data
Chat feature data lands in the same warehouse as SERP feature tracking, prompt research, and citation analytics. A single view can put a topic's behaviour on Google beside its behaviour in AI answers, which is where most of the useful arguments start.
Available in the application, through the API, and in the Search Intelligence BigQuery warehouse for customers with that add-on. Related: SERP Features, Citation Analytics, and Prompt Research.
See it with your own data.
30-minute demo. We'll run it on your domain - no prep required.