Made expert statistical analysis self-serve, for every customer
Move one of four. I brought analysis that experts did by hand into the product, matched it to their results, and designed it so any manager could act on it.
- Company
- Hive HR
- Year
- 2024
- Role
- Product Manager (solo)
- Team
- Data team, engineering
- Product strategy
- UX
- Data products
At a glance
- The problem
- Analysis that showed customers what really drives employee engagement was done by hand, so only customers paying for expert services got it.
- My role
- Product owner. I defined how it worked in the product, held it to the experts' results, and designed the experience.
- The result
- Instant, consistent analysis for every customer, trusted enough to stand behind. It became the foundation for the next three moves.
The problem
Key driver analysis shows what actually moves engagement, not just which scores are high or low. It was one of our biggest selling points, but our data team ran it by hand in specialist software, so most customers never saw it.
There was a real risk in automating it. Customers had trusted these results for years. If the product gave different answers from the same data, we’d be recommending different actions, and that loses trust for good.
What I did
No outside tool could match the quality we needed, so we built our own analysis engine. I added a simpler method alongside the main one, so small organisations still got reliable answers instead of nothing, or something misleading.
Then I designed for managers, not analysts. A line manager doesn’t care about statistics, so the whole experience rests on one idea:
If this score improves, your engagement score is likely to improve too.
A single grid shows how strongly each factor matters against how well it’s doing now. Where it lands tells you what to do: prioritise, maintain, celebrate or review.
The result
- Instant analysis for every customer, not just those paying for services.
- Trust held, because results matched the experts’ closely enough for teams to stand behind them.
- The data team stopped doing it by hand, and we cut specialist software costs.
- In testing, users preferred it to competitors’ chart-heavy versions.
It also unlocked what came next. Hive Actions and Auto Insights both build on this engine.
“So user-friendly for managers.” A housing association customer