From Voice of Customer insights to a shipped feature in 8 weeks.

Customers were losing work in an enterprise GenAI tool. The drop-offs showed up in telemetry. The frustration showed up in support tickets. The anxiety showed up in drop-in sessions.

The case for autosave came from triangulating three independent channels.

The instinct in product conversations was to frame features by what they add. This case was built in reverse — quantifying what inaction was costing, in terms the business already understood: billable time. Using rate card data and session telemetry, each abandoned session could be given a rough cost. Aggregate across the user base and the number became hard to ignore.

But there was a harder problem. The product team was building toward the AI-powered capabilities that were part of the broader platform vision. Getting autosave prioritised meant reframing it as an adoption prerequisite. Simply put, I argued that users can't realise value from a feature they never reach.

The case moved through three conversations — product owner, senior leadership, and the cross-discipline delivery team. Each needed a different framing of the same core argument.

The product owner needed the commercial model: quantified waste in a cost-conscious delivery is harder to ignore than a user rejection. Leadership needed the adoption risk framing stripped of feature detail. The delivery team needed the drop-in insights translated into acceptance criteria — ensuring what got built matched the user's mental model of when and how saves happened.

Post-release telemetry confirmed the pattern. Session abandonment at the 18–35 minute threshold fell ~40%. Duplicate deliverables dropped. Average session length increased.

The most telling signal was from a follow-up drop-in session: "It sounds small but it actually changed how I use it. I just leave it open now."

What's not on this page: the client, the product, the exact numbers. I can talk through some details in conversation.