AI skills
agile-discovery
Use when asked to run the discovery track of dual-track agile — validating what to build before building it, via problem identification, user personas, and prototyping — as the track that runs alongside, not before, delivery (see agile-delivery), and feeds product-management's broader discipline.
Discovery is the process of identifying, defining, and validating what a product or feature should actually be — before committing engineering effort to building it. In dual-track agile, discovery runs as a continuous track alongside the delivery track (see [[agile-delivery]]), rather than as a one-time upfront phase that finishes before delivery starts: discovery is always a sprint or two ahead, validating what delivery will build next.
The discovery process
- Identify the problem — what does this solve, really? Market research, customer conversations, and analysis of existing solutions ground this in reality rather than internal assumption (see [[voice-of-the-customer]] for the customer-feedback side of this).
- Define user personas — fictional but research-grounded representations of the target audience, clarifying whose needs and goals the solution actually has to serve.
- Brainstorm potential solutions — generate options against the personas' needs (see [[ideation]], [[brainstorming]]), evaluating each for feasibility, desirability, and viability — the three-part test commonly used to judge whether an idea is actually worth building.
- Prioritize features — what do users actually want, using research or a prioritization framework such as MoSCoW.
- Create prototypes — low-fidelity wireframes or a working high-fidelity prototype, tested with real potential customers before committing to full build effort.
- Iterate — gather feedback on usability, functionality, and appeal, and refine based on what's actually learned, not just the first prototype's reception.
- Develop a product roadmap — once the feature set is validated, translate it into milestones, timelines, and priorities feeding delivery.
Why discovery and delivery run in parallel, not sequentially
A strict "discovery phase, then delivery phase" split risks two failure modes: delivery sitting idle waiting on discovery to fully finish, or discovery rushed to unblock delivery before ideas are actually validated. Running discovery continuously, staying a step ahead of delivery, means delivery always has validated work ready to build while discovery keeps investigating what comes next — validated ideas flow from discovery into delivery's backlog continuously rather than in one large upfront batch.
Relationship to product management
Discovery is the concrete, ongoing activity that [[product-management]]'s "discovery vs. delivery" framing describes at a more general level — this skill covers the specific process steps; product management covers the broader strategic context (prioritization frameworks, success metrics) discovery operates within.
Common pitfalls
- Treating discovery as a one-time phase before "real work" starts — defeats dual-track agile's whole premise; discovery should be continuous, staying ahead of delivery rather than finishing before it.
- Skipping prototyping and validation — building based on internal assumption rather than tested prototypes reintroduces exactly the risk discovery exists to reduce.
- Personas built from assumption, not research — see [[user-centred-design]]'s identical warning; an unvalidated persona gives false confidence rather than real insight.
- Discovery that never feeds delivery — validated ideas that stay in discovery indefinitely, never translated into a roadmap delivery can act on, produce research without outcome.
Learn more
- [[agile-delivery]] for the parallel delivery track discovery feeds.
- [[product-management]] for the broader strategic discipline discovery operates within.
- [[voice-of-the-customer]], [[user-centred-design]] for the customer-research methods discovery draws on.