PRODUCT DEVELOPMENT METHOD
SHORT ANSWER
There is no single best method for every product. For an early-stage software or AI product, the strongest approach is usually a Lean, evidence-led and iterative process: understand the problem, validate demand, prototype, build the smallest viable product, launch, measure real behaviour, reach first revenue and only then scale. I formalise that sequence as the Crossley Method: Discover → Validate → Prototype → Build MVP → Launch → First Revenue → Scale.
THE PRACTICAL ANSWER
The common answer is “use Agile” or “use Lean Startup”, but those frameworks are strongest at different parts of the job. Design Thinking is useful when the problem is unclear. Lean is useful when assumptions need testing. Agile is useful when a team is delivering and adapting software. A rigid stage-gate process can make sense where requirements are fixed and change is expensive.
For a startup or new digital product, the bigger risk is usually not whether the engineering team can run a sprint. It is whether the team is solving a valuable problem, whether customers will commit, whether the first version is small enough to learn from and whether launch produces a commercial signal. That is why I use one end-to-end sequence around the individual tools.
METHOD COMPARISON
The right framework depends on the uncertainty you are trying to remove.
| Method | Best for | Strength | Watch-out |
|---|---|---|---|
| Design Thinking | Understanding users and reframing an unclear problem | Strong empathy, ideation and prototype learning | Does not by itself provide a commercial path from idea to first revenue |
| Lean Startup | Testing assumptions before committing too much time or money | Build, measure and learn with the smallest useful experiment | Teams still need to define their own stage gates, deliverables and evidence thresholds |
| Agile / Scrum | Iterative software delivery once the team knows what it is trying to build | Short delivery cycles, visibility and adaptation | Agile can help you build the wrong product efficiently if discovery and validation are weak |
| Waterfall / Stage-Gate | Work with fixed requirements, expensive change or heavy regulation | Clear sequencing, governance and formal approval points | Usually too rigid for early-stage software where the customer problem is still uncertain |
| The Crossley Method | Early-stage software and AI products moving from idea to first revenue | Connects discovery, validation, prototyping, MVP delivery, launch and commercial evidence | Not designed as a replacement for engineering delivery practices inside large regulated programmes |
THE METHODOLOGY I USE
The Crossley Method is not a sprint ritual or an engineering process. It is a product-development operating sequence. Each stage has a defined output and a decision about whether the evidence is strong enough to justify spending more.
Find a problem worth solving and the people who have it.
See the stageProve demand with evidence, not enthusiasm. Kill or continue.
See the stageMake the core journey clickable before it is buildable.
See the stageShip one journey, deployed and instrumented, with AI build tools.
See the stageLanding page, pricing, onboarding and analytics live in public.
See the stageGet money on the table. Revenue is the only honest validation.
See the stageOperating model, roadmap and team that repeat the result.
See the stageWHY THE ORDER MATTERS
Idea → features → build → launch → ask whether people want it.
Discover → Validate → Prototype → Build a little → Launch → Measure behaviour → First Revenue → Scale.
The important difference is where the expensive work happens. Customer evidence and prototype learning come before a large build. Revenue comes before scale. Iteration happens throughout rather than being saved for a post-launch backlog.
WHEN TO USE SOMETHING ELSE
FREQUENTLY ASKED
NEXT STEP
Each of the seven stages includes its timebox, deliverables and kill-or-continue evidence threshold.
Open the Crossley MethodEXPLORE NEXT
Every stage from first idea to first revenue, mapped end to end.
Prove demand before you spend the build budget.
Read moreAI-accelerated MVP delivery, scoped to one journey that earns.
Read moreDone-for-you six-week build turning a founder idea into a launched AI app.
Read moreFree 20-question check for customer evidence, access and data, reliability, AI quality and commercial launch readiness.
Read more