03 — Form Testable Hypotheses
Purpose: Convert assumptions into measurable hypotheses that can be validated.
Outcome
Decision-relevant hypotheses with explicit expected outcomes.
This stage is where the team makes the shift from belief to test design.
Time to complete
30-60 minutes.
Inputs
Prioritized assumptions from Step 02.
Available channels, users, and test constraints.
Steps in SwiftCNS
Convert each priority assumption into one hypothesis.
Add measurable expected outcome and timeframe.
Define what invalidation looks like in advance.
Select hypotheses for immediate experiment planning.
Why this stage matters
Assumptions tell the team what it is uncertain about. Hypotheses tell the team how it will know whether that uncertainty is becoming clearer.
This stage matters because weak hypotheses create vague experiments. If the expected outcome is not measurable, the team can run the test and still argue about what the result means afterward.
Role lenses
Startup: keep hypotheses simple and high-signal.
Program manager: enforce consistency in hypothesis quality across teams.
Mentor: check for testability and weak wording.
What strong output looks like
A strong hypothesis:
is tied to one assumption,
has a measurable expected outcome,
can be tested in a realistic timeframe,
helps the team decide what to do next.
Weak vs strong pattern
Weak
hypothesis is broad or inspirational,
expected outcome is unclear,
no visible invalidation condition,
test would not meaningfully affect a decision.
Strong
hypothesis is narrow and testable,
outcome is measurable,
success and failure criteria are visible,
result would change confidence in a meaningful way.
Outputs
Hypotheses ready for experiment design.
Success and failure criteria for each hypothesis.
Definition of done
Hypotheses are testable, measurable, and time-bounded.
Team agrees which hypotheses move to experiments first.
Common failure mode
The usual trap here is writing hypotheses that sound sharp but are still too broad to guide a real experiment. If the team cannot tell what evidence would invalidate the statement, the hypothesis is not finished yet.
If blocked
Use Quality Standards to tighten measurable outcomes.
Next step
Continue to 04 — Design & Run Experiments.
Last updated