Build-Measure-Learn: The Waste-Elimination Lens
Treat every unvalidated feature and every dollar spent as inventory you must justify or cut.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 80%
Eric Ries's core Lean Startup mechanism reframes product development through a manufacturing lens: designed-but-undeployed features are 'work in progress inventory,' and untested, unvalidated assumptions are liabilities, not assets. The discipline forces founders to ship in small batches (single-piece flow), measure real customer behavior, and only keep spending that generates validated learning. Ries developed this while running engineering at IMVU, shipping to production dozens of times a day, and later used it to coach founders during the 2008 financial crisis who were cutting costs blindly instead of asking whether each dollar produced learning. The framework's power is that it applies equally in growth mode and in crisis mode, because the test is always the same: does this spend teach us something critical right now?
Origin
Developed while Ries ran engineering at IMVU, translating lean manufacturing concepts (single-piece flow, inventory-as-waste) directly onto software development; later generalized into a cost-cutting lens during the 2008 financial crisis.
Core principles
- 01Untested, unvalidated assumptions are inventory and are bad, not good
- 02Designed but undeployed features are work-in-progress and a liability
- 03Ship in the smallest possible batches (single-piece flow) to surface learning fast
- 04Every dollar of spend should be evaluated by whether it produces critical learning
How to run it
- 1
Identify the riskiest assumption
Before building anything, name the assumption whose failure would kill the business or the feature.
Pro tip Write the assumption down as a falsifiable statement, not a vision statement.
- 2
Build the smallest possible test
Ship a minimum viable version using single-piece flow rather than batching many features together.
Watch out Avoid letting designed-but-undeployed features pile up; they are inventory, not progress.
- 3
Deploy continuously
Push to production frequently (IMVU shipped up to 40 times a day) so learning arrives in near real time.
- 4
Measure actual behavior
Collect real usage data rather than relying on internal opinion or theory about what customers want.
- 5
Audit every dollar of spend
For any expense, ask whether it is helping you learn something critical about the company right now.
Pro tip Use this test in both boom times and crisis/cost-cutting times — the test doesn't change.
Watch out Cutting cost for its own sake (e.g., renegotiating furniture leases) misses the point if it isn't tied to learning.
- 6
Decide: persevere or pivot
Use the validated (or invalidated) learning to decide whether to keep the current approach or change direction.
In the wild
Ries explained to skeptical VC-referred management teams that his team shipped software to production up to 40 times a day, treating every unvalidated feature as inventory to be tested quickly rather than a batch to be perfected before launch.
→ Teams initially rejected the idea as impossible at their scale, but the practice became the seed of what Ries later formalized as the Lean Startup method.
Founders called Ries for help cutting burn rate (getting out of office leases, selling furniture). He redirected them to ask whether each dollar of spend produced critical learning, not just whether it looked frugal.
→ Reframed cost-cutting as a learning discipline rather than panic-driven austerity.
Common mistakes
Confusing built features with progress
Designed-but-undeployed features feel like output but are actually unsold inventory sitting as risk on the balance sheet of the company.
Cutting costs without asking the learning question
Panicked cost-cutting (like furniture repossession) misses the actual point: only cut spend that isn't producing validated learning.
Is it for you?
Best for
Early-stage product teams and founders trying to decide what to build next or what spending to cut.
Not ideal for
Situations requiring a single big-bang launch for regulatory, safety, or brand reasons where iterative testing in market isn't feasible.
From the transcript
“Untested, unvalidated assumptions are inventory and are bad, not good.”
“the build measure learn feedback loop. The reason it's important is we can analyze every dollar you spend and we can ask ourselves is that…”
From the episode
The Lean Startup and the Long-Term Stock Exchange (with Eric Ries)
Eric Ries