Analytics Lenses
Slice every analytics number by segment, intent, journey stage, topic, tag, prompt group, or Query Universe orbit — composable, optional filters with honest sample-size disclosure.
Overview
The lens bar sits at the top of analytics pages and slices every number on the page to just the prompts that match your filter. The familiar segment pills (All / Branded / Non-Branded / Competitor) are one lens dimension — six more are available as compact optional chips.
A lens applies everywhere at once: the same filter works on the dashboard, Analytics, Trends, Sentiment, Competitors, Insights, and more — and the identical filter is accepted by the REST API, MCP tools, and report generation, so a number you see in the UI can be reproduced anywhere.
The seven dimensions
| Dimension | What it slices by | Where the values come from |
|---|---|---|
| Segment | Branded / Non-Branded / Competitor | Auto-classified per prompt (overridable) |
| Intent | Commercial-intent type (Informational, Commercial, Transactional, Hybrid…) | Auto-classified per prompt |
| Journey | Buyer-journey stage (awareness → decision) | Auto-classified per prompt |
| Topic | Topic cluster | The prompt's Topic field |
| Tag | Your custom tags (campaigns, products, teams…) | Tags you set on prompts — up to 10 per prompt; the distinct-tag ceiling depends on your plan |
| Group | A saved Prompt Group | Groups you curate on the Prompts page |
| Orbit | Query Universe orbit group (what-we-do, how-you-qualify, whom-we-serve) | Prompt links to Query Universe building blocks |
Every filter is optional
Every dimension is independent and defaults to All. Picking a single lens — say, Intent: Commercial — is the primary path; no filter ever requires another. Combining lenses intersects them: Intent: Commercial + Orbit: what-we-do means commercial-intent prompts inside that orbit.
A chip only appears when its dimension is actually usable on your project: no Orbit chip without a Query Universe, no Group chip without prompt groups, no Tag chip before your first tag, no Topic chip unless your prompts span more than one topic.
Honesty under filters
Filtering makes samples smaller, and the platform says so instead of letting a thin slice masquerade as a solid read:
- The bar shows "N prompts in filter" so you always know the slice size. An empty slice shows an explicit "No prompts match this filter — widen or clear it" state, never silently blank charts.
- Sliced numbers carry the same sample-size warnings as everywhere else: below 30 runs a rate is directional, below 10 it is low-confidence.
- A few metrics count the whole project because their data isn't stored per prompt — Active Anomalies, Model Drift, and the Actions board. While a lens is active they display an "All prompts" badge rather than pretending to filter.
- Some surfaces are structurally unfilterable and stay lens-free by design: GA4/GSC fusion, revenue attribution, and the Citations source directory.
- If a filter ever fails to resolve, the page shows an empty state — filtered views never silently fall back to unfiltered data.
Shareable views
Lenses live in the URL (for example ?intent=Commercial&orbit=what-we-do), so the address bar reproduces the exact sliced view for anyone with project access. Browser back/forward walks your filter history, and scheduled reports can be pinned to a lens so a stakeholder's weekly PDF covers exactly their slice.
Tip: Filter to Intent: Commercial or a decision-stage Journey to see your AI visibility on the prompts closest to revenue — that's the slice where visibility gains convert.