Interactive demo — sample company dataEngine running

Learning history

Every experiment ends in a learning — validated or invalidated — and each one updates the company model.

Validated learnings
23across all cycles
Invalidated beliefs
4caught before scaling
Avg. confidence lift
+14%per validated learning
Evidence collected
1,842signals feeding the model

July 2026

  • Validated

    High-intent channels (G2, branded search) convert trials at 4x paid social.

    312 trials across 9 weeks · Channel holdout test

    Confidence54%88%
  • Invalidated

    In-platform CAC predicts downstream customer quality.

    90-day churn follow-up, 2 cohorts · Cohort quality model

    Confidence61%22%
  • Validated

    Cost-saving messaging outperforms CX messaging for finance buyers.

    2.4% vs 1.1% CTR, 41k impressions · Ads + search experiment

    Confidence48%82%

June 2026

  • Validated

    Personalized outbound angles more than double reply rates at current volume.

    7.1% reply rate across 8 variants · Outbound experiment

    Confidence45%71%
  • Invalidated

    “Delight customers” positioning resonates with mid-market buyers.

    CTR 1.1% vs 2.4% control · Paid social ad set

    Confidence52%28%
  • Validated

    Teammate invites during onboarding are the strongest activation lever.

    61% of activated accounts invited a teammate · Onboarding prompt test

    Confidence39%61%

May 2026

  • Validated

    Product-led activation drives the highest quality pipeline.

    PLG conversion up 24%, CAC down 18% · Funnel analysis

    Confidence50%78%
  • Invalidated

    Longer trials increase conversion for self-serve signups.

    14-day vs 28-day trial split · Trial length experiment

    Confidence58%31%