The learning spiral
Phase 5 is not the end of the methodology. It's the transition from a linear cycle to a learning spiral. Each turn of the spiral produces a more precise Context Document, a solution more faithful to the real problem, and a deeper understanding of the market.
The distinction between cycle and spiral
The distinction is fundamental: a cycle returns to the same point. A spiral returns to a higher point.
This has a direct practical consequence. The question is not "when does Phase 5 end?" The right question is "what did I learn in this turn that I didn't know at the start?" If the answer is "nothing," there is no spiral. There is a sterile cycle.
Iterating is not repeating. Each iteration must incorporate new information. If you return to the same point without having learned something new, you're not iterating — you're going in circles. An iteration without learning is a cost with no return.
The phase return map
The learning spiral has a structure that determines which phase the team returns to with each learning:
| Type of learning | What the market says | Returns to which phase | Example |
|---|---|---|---|
| The problem was poorly defined | Users don't adopt the solution because it doesn't solve what hurts them most. The real pain is different from what was documented. | Phase 1 — Problem Phase | The inventory dashboard isn't used. Follow-up interviews reveal the real problem wasn't visibility but trust in the data. |
| The solution doesn't fit | The problem is well defined. The solution produces the expected result but adoption friction is too high. | Phase 2 — Solution Phase | The alert system works but managers silence them because they arrive during peak operational load. |
| The context needs adjustment | User behavior is consistent with the solution. KPIs are moving but not at the expected pace. | Phase 3 — Context Phase | The predictive model works but its threshold generates too many false positives. The Rules need recalibration. |
| Implementation optimizations | KPIs are moving in the right direction. Users adopt the solution. There are performance improvements or new integrations that would expand value. | Phase 4 — AI Build Phase directly | Competitors have a logistics system integration that would add value. It's added as a new set of Story Files. |
This map is not a rigid decision tree. It's a thinking framework. The team that learns to read the type of signal the market is sending — definition problems vs. design problems vs. calibration problems — is the team that iterates with precision instead of iterating blindly.
How the spiral generates cumulative precision
Each turn of the spiral produces a Context Document that could not have been created without the previous turn. Phase 1 hypotheses are confirmed or refuted. Phase 3 assumptions are calibrated with real data. Phase 4 decisions are evaluated with observed behavior.
A project that has been in Phase 5 for twelve months has a Context Document that no amount of initial discovery could have produced. Not because the discovery was insufficient, but because the information the market produces when using the system is of a quality that no pre-launch process can match.
If after three iteration cycles the team cannot articulate what it learned that it didn't know before, the spiral isn't working. Likely cause: implementation changes are being made without updating the Context Document, or activity (features delivered) is being measured instead of learning (assumptions verified).