Does anything go live without my approval?

No. Every variant and audience is staged for review. You approve, then publish.

What data does it need?

Tracked conversion events and a connected experiment audience. For recommendations, a product catalog uploaded to Intempt.

Yes. The segment that ran the experiment becomes the personalization audience — no rebuilding, no export.

Experience Optimizer · Part of Blu

For growth and product teams. Runs structured CRO cycles, personalizes every surface by segment, and reviews your entire testing program quarterly.

Click Invoke on any skill to run it live in Blu.

/cro-program

8 ACTIONS

Baseline → drop-off → ICE scoring → experiment → analysis → next test. A program.

Builds: Funnel Report · ICE Scorecard · A/B Experiment +3 more

/personalization-recs

7 ACTIONS

Right algorithm per surface, real-time audiences, server-side and client-side, attribution dashboard.

Builds: Recommendation Feeds · Audience Configs · Personalization Rules +1 more

/testing-retrospective

5 ACTIONS

Win rate, portfolio stats, coverage gaps, next quarter roadmap, executive summary.

Builds: Win Rate Analysis · Portfolio Stats · Coverage Gap Report +2 more

/vip-program

6 ACTIONS

LTV-based tiers, exclusive experiences, holdout experiment, quarterly loyalty dashboard.

Builds: VIP Segments · Tier Definitions · Exclusive Experiences +2 more

A consistent pattern across every skill.

Blu builds your funnel, ranks each step by ICE score, and surfaces the highest-impact opportunity to test.

Hypothesis, variants, primary metric, guardrail metric, and sample size, all defined before a single pixel moves.

Blu tracks significance and confidence intervals in real time. No spreadsheet. No manual calculation.

When results are ready, Blu recommends ship or kill with the reasoning, then designs the next test from what you learned.

Experience Optimizer is one of 9. Each agent owns a piece of the GTM motion, from research and outreach to lifecycle and revenue ops.