Skip to main content

Tavism
All enterprise casesInternal Experimentation

Low-Cost CSR & Volunteering Experimentation

Hewlett-Packard — Enterprise Experimentation

Do not buy the global portal first. Run small volunteering tests until participation and ops numbers earn the software spend.

Case overview

The challenge

Executive Caroline Barlerin set out to launch global employee volunteering programs at Hewlett-Packard. The default path was a worldwide CSR portal. That bet would lock budget into software before anyone knew whether employees would show up—or which ops bottlenecks would break the program.

Small-batch testing

Low-cost experiments with small cohorts

Instead of launching a massive worldwide portal, the team ran low-cost experiments with small employee cohorts. Each batch tested real signup, matching, and follow-through—cheap enough to kill or change before a global rollout.

Validated learning

Measure participation before the software buy

They measured participation rates and operational bottlenecks first. Those numbers decided whether a global software rollout was worth funding—and which process fixes had to land before the tech.

Results

Record engagement, less corporate waste

The small-batch path reached record employee engagement while cutting waste that a premature portal would have baked in. Learning came before the large capital outlay—not after.

Portal betCohort proof

Capital timing after validated volunteering demand

For executives

Key takeaways for executives

How to run CSR and internal programs as enterprise experiments.

  1. Treat a global volunteering portal as a capital bet—not a starting assumption.
  2. Test with small employee cohorts until participation and ops bottlenecks are real numbers.
  3. Fund the worldwide software rollout only after cohort learning earns it.
  4. Record engagement with less waste is the win—not the earliest enterprise tool purchase.

Ready to test before the enterprise rollout?

Book 30 minutes. Bring the internal program stuck behind a big software buy and the metrics that would kill or fund it.

Run Enterprise Experiments