The Hypothesis

Can a dashboard generate itself — no configuration, no widget selection — just by knowing what campaign is running and what outcomes matter?

The Concept

Every dashboard tool requires setup: select metrics, choose visualisations, configure data sources, arrange widgets. This experiment eliminates all of that. When a campaign goes live — landing page traffic, email test, ad launch, lifecycle flow — the dashboard auto-generates itself. The system knows the campaign type, the defined outcome goals, and what data is flowing through the feedback loop. It dynamically assembles the right layout, metrics, and charts for that context. An AI narration layer explains what's happening in plain English alongside every chart.

Auto-generating campaign dashboards

The hypothesis

Can a dashboard generate itself — no configuration, no widget selection — just by knowing what campaign is running and what outcomes matter?

The concept

Every dashboard tool requires setup: select metrics, choose visualisations, configure data sources, arrange widgets. This experiment eliminates all of that. When a campaign goes live — landing page traffic, email test, ad launch, lifecycle flow — the dashboard auto-generates itself. The system knows the campaign type, the defined outcome goals, and what data is flowing through the feedback loop. It dynamically assembles the right layout, metrics, and charts for that context. An AI narration layer explains what’s happening in plain English alongside every chart.

How it works

1. Campaign goes live — landing page, email, ad, or lifecycle flow activates
2. System detects campaign type — determines what metrics and views are relevant
3. Auto-assemble dashboard — outcome-first layout: business goal at top, supporting metrics below
4. AI narrates in real time — “Variant B outperforming by 18% after 1,200 sessions. ~300 more needed for significance.”
5. Chat-queryable — “Why is the Tuesday cohort converting lower?” → data-backed answer

No configuration. No widget selection. The dashboard appears when the campaign starts.

What it explores

• Can auto-generated dashboards match the usefulness of manually configured ones?
• Does outcome-first layout (business goal at top) change how teams make decisions?
• How accurate and useful is AI narration compared to raw data interpretation?
• Does chat-queryable data access reduce time-to-insight vs. traditional filtering?
• Can the system reliably choose the right visualisation type for each metric?

What we found

• Time-to-first-insight dropped from hours (manual dashboard setup) to seconds (auto-boot)
• AI narration was reported as more useful than charts — it surfaced insights teams would have missed scanning data alone
• Outcome-first layout shifted team conversations from “what happened” to “what should we do”
• Chat queries handled 70% of ad-hoc data questions without anyone touching a filter

Learnings

• Auto-generating layout by campaign type covers 80% of use cases — the other 20% need a lightweight customisation escape hatch, not a full builder.

• Raw charts without narrative context are often misinterpreted — every visualisation should ship with a plain-English explanation.

• Chat-queryable access works for ad-hoc questions but scheduled reporting still needs traditional exports — different cadences need different interfaces.

• Dashboard auto-boot builds trust in autonomous systems — when teams can see what the system is doing, they let it do more.

Where it goes next

This is becoming a core component of Lens. The narration layer in particular — the idea that every chart should come with a plain-English explanation of what it means — is something we’re applying across all our reporting surfaces.