Most failed features weren't badly built — they were never validated. Discovery is how teams find that out for a few days of interviews instead of a few months of engineering.
1Discovery Is Weekly, Not Quarterly
A single round of user interviews at the start of a project isn't discovery — it's a kickoff ritual. Continuous discovery means the product trio (PM, design, engineering lead) talks to customers every week, tied to a specific outcome the team is trying to move, so assumptions get tested continuously as the product evolves rather than once at the very start.
2Mapping Opportunities Before Solutions
It's tempting to jump straight from a customer complaint to a backlog ticket. The Opportunity Solution Tree forces an intermediate step: cluster what you're hearing into distinct opportunities, weigh them against the outcome you're driving, and only then generate multiple candidate solutions per opportunity so you're not betting everything on the first idea that came to mind.
3Step-by-Step Breakdown
Introduction. Product Discovery is the ongoing work of figuring out whether an idea is worth building — before it becomes a sprint commitment. It answers four questions: will the user choose it, can we build it, will our business benefit, and will stakeholders support it.
Continuous Discovery Habits. Teresa Torres' model calls for weekly touchpoints with customers, run by the team building the product, tied to a specific product outcome. One interview six months ago doesn't count — discovery is a habit, not a phase you finish before 'real work' starts.
The Opportunity Solution Tree. An Opportunity Solution Tree maps a desired outcome to the customer opportunities (pain points, unmet needs) that drive it, then to candidate solutions, then to the smallest experiment that tests each one. It keeps teams from jumping straight from 'idea' to 'roadmap item'.
Knowledge Check. In a discovery interview, why is 'Would you use a feature that lets you export your data to PDF?' a weaker question than 'Walk me through the last time you needed to share this data with someone'?
- →It asks about past, real behavior instead of a hypothetical future preference, which people are notoriously bad at predicting
- →It is a shorter sentence, and interview questions should always be kept under 10 words
Summary. Discovery isn't a gate before 'real' product work — it's the weekly habit of testing assumptions with real customers so the team enters delivery already knowing the problem is worth solving.
