Watch the workaround
When: Usage data shows the product being used in ways nobody designed for. Support tickets describe elaborate hacks. Customers export to spreadsheets, chain tools together, or misuse a feature at scale, and the team's instinct is to call it edge-case behaviour.
The action: Go and watch the workaround happen where it happens. Treat compensating behaviour as the most honest signal of an underserved job, because people improvising under their own effort are demonstrating demand no survey can fake. Trace what job the hack serves before deciding whether to absorb it into the product.
Forces: Observation is expensive per data point against the abundance of cheap telemetry, and it requires customers willing to be watched. The pull to explain workarounds away as user error is strong because taking them seriously implicates the roadmap. Some hacks really are noise, and telling demand from idiosyncrasy is judgement work.
Signals: Within weeks, the team can describe at least one workaround in terms of the job it serves rather than as misuse. A discovery bet is opened, or explicitly declined, on evidence from observed behaviour rather than stated requests.
When not to: When the workaround belongs to a handful of power users whose job does not generalise, since paving every desire path builds a maze. When observation would breach trust or confidentiality that matters more than the insight.
Instance: Intuit's follow-me-home practice, through which watching Quicken customers revealed a large share running businesses on a personal finance tool, a workaround that became QuickBooks.
Licensed under CC BY-SA 4.0.