How we ran the programme
Ninety design partners across mid-market and enterprise. Each one paired with a named engineer or PM. Weekly issue triage. Monthly group calls. A shared issue tracker the partners could read. The discipline was that every issue had a named owner and a documented status.
We were honest about what we could and could not do. Design partners respect 'no, and here is why' more than 'maybe'.
The five biggest product changes
Custom screening parameters on top of the six defaults. Microsoft Teams interview rooms (alongside Google Meet). One-click posting to LinkedIn, Naukri, and Indeed. The nine-tab candidate workspace replacing the older modal. Drag-drop feedback form builder.
All five came from design partners. The behind-the-scenes post covers the screening engine specifically; the rest landed across the platform.
The feature we walked back
We shipped a smart-default that auto-rejected candidates whose score fell below a configurable threshold. Two design partners loved it. Most disliked the loss of human judgment. We removed the auto-reject and kept the threshold-based highlight, which is the diff that mattered.
The lesson: convenience features that erode visible judgment are usually the wrong call, even when the model is good.
What we kept after the programme
The named-owner discipline. Every product issue has a single owner who responds within a working day. Customers, not just design partners, see the same response cadence. Talk to us if you want to see how the issue tracker looks; we are happy to share.