Product Discovery
The work of reducing uncertainty about which customer problems and product solutions are valuable, usable, feasible, and viable before scaling.
Download Markdown (.md)Soroush’s Product Wiki · https://wiki.soroush.uk/topic/product-discovery/
01What it means
Product discovery is a collaborative learning process used to decide what to build and why. A cross-functional team studies customers and context, frames opportunities, explores alternatives, and tests critical risks before committing substantial delivery effort, while continuing to learn after release.[1][2]
- 01Set outcome + decisionConstraints and important unknowns
- 02Gather evidenceCustomers, behavior, market, operations, technology
- 03Frame opportunityPrioritize problems, not feature lists
- 04Explore alternativesProduct, design, engineering perspectives
- 05Test + decideProceed, change, learn more, or stop
Discovery informs what to build and why; it continues alongside delivery instead of becoming a one-time phase.
02Why it matters
Delivery efficiency is wasteful when a team is confidently building the wrong thing. Discovery improves the quality of product decisions by testing value, usability, feasibility, and business viability early and by keeping evidence close to the people who will design and build the product.
03When to use it
Use discovery for new products, major capabilities, uncertain customer problems, risky business changes, and meaningful redesigns. Scale the effort to the uncertainty and cost of being wrong; small reversible changes should not inherit a heavyweight ceremony.
04How it works
- Clarify the decision, desired outcome, constraints, and important unknowns.
- Gather evidence from customers, behavior, markets, operations, and technical systems.
- Frame and prioritize opportunities rather than beginning with a feature list.
- Explore multiple solutions with product, design, and engineering perspectives.
- Test the riskiest assumptions using the smallest credible evidence.
- Decide whether to proceed, change direction, gather more evidence, or stop.
05A practical example
A logistics platform considers automated delivery estimates. The trio interviews dispatchers, reviews late-shipment data, prototypes confidence ranges, and runs a technical spike on data quality. It learns that exception explanations create more trust than a single precise time and adjusts the concept before delivery.
06Where it breaks
Discovery cannot remove uncertainty, prove future demand, or replace judgment. Poor samples and leading tests create false confidence, while endless research can delay decisions. Discovery also fails when teams lack authority to act on evidence or exclude delivery and operational realities.
Common mistakes
- Running discovery as a design phase handed off to engineering.
- Seeking validation of a chosen solution rather than comparing options.
- Using research volume as a substitute for a decision and learning threshold.
07Sources
- Product DiscoverySilicon Valley Product Group · Practitioner articleProduct Discovery
- The Origin of Product DiscoverySilicon Valley Product Group · Practitioner historyThe Origin of Product Discovery