Skip to main content

We Tested Everything. We Validated Nothing.

25-minute Talk

Comprehensive testing, zero technical issues. Dozen million users, flat business results. Explore what happens when testing validates execution brilliantly but misses strategic value.

Virtual Pass session

Timetable

10:45 a.m. – 11:30 a.m. Thursday 19th

Room

Room E1 - Track 4: Talks

Leadership Machine Learning (ML) Test Management

Audience

Testers and QA professionals who need to bridge technical validation and business value.

Key-Learnings

  • The Testing Paradox: Perfect technical testing cannot save a product solving the wrong problem. "Does it work?" is different from "does it matter?"
  • Metrics vs Meaning: High engagement + zero bugs looked successful until we discovered we tested activity, not intent.
  • Testing Data Quality = Testing Product Quality: For ML/AI systems, testing data signals is as critical as testing code.

Introduction (2 mins)
The setup and hook: Perfect testing, all green, but flat business performance.

What We Tested Right (5 mins)
Our comprehensive testing approach - what passed and why we trusted it.

What Testing Missed (8 mins)
The gap discovery: technical validation ≠ business value. How we discovered users clicked from confusion (tested activity, not intent) and perfect algorithms on poor data signals (tested code, not data quality).

How We Tested the Fix (7 mins)
The immediate question: What should we test NOW to validate the fix?
What changed in our testing and the result:

  • De-Prio: Testing engagement (clicks, time-on-page)
  • Started: Testing conversion funnels (click → browse → add → purchase)
  • Added: Data quality validation (are signals showing intent or confusion?)
  • Built: Fast feedback loops and human intervention

Key Takeaways (3 mins)
The framework: Recognition (spot the gap) + Action (test differently). Simple questions to build into your testing strategy now, before the failure happens.

Related Sessions

There are currently no related sessions listed. Please check back once the program is officially released.