Hypothesis Creation

Definition

A hypothesis is a testable, falsifiable statement about what you believe to be true — about a user, a problem, or a solution — that you design an experiment to confirm or refute.

Why Hypotheses Matter in Discovery

Without an explicit hypothesis, teams "test" things without a clear pass/fail line, then interpret ambiguous results however suits their preference (confirmation bias). A good hypothesis forces clarity before you look at any data.

Hypothesis Statement Format

"We believe that [doing X] for [user segment] will result in [outcome], measured by [metric]. We will know this is true when we see [specific signal/threshold]."

Good vs. Bad Hypotheses

• Bad: "Users will like the new design." (Not falsifiable, no metric.)
• Good: "We believe redesigning the checkout flow will reduce cart abandonment for mobile users by at least 10%, measured by funnel completion rate over 2 weeks."

Categorizing Assumptions

Every solution rests on a stack of assumptions. Sorting them by type helps you find the riskiest one to test first.

• Desirability assumptions — do users want this? (value risk)
• Usability assumptions — can users figure out how to use it?
• Feasibility assumptions — can we technically build it within constraints?
• Viability assumptions — does it work financially/legally/strategically for the business?

Leap-of-Faith Assumptions

Among all assumptions, the leap-of-faith assumption is the one that, if false, breaks the entire idea. Identify it explicitly — it's the one worth testing first, even if it's uncomfortable to question.

Assumption Stacking

Solutions often have assumptions stacked on assumptions (e.g., "users want X" assumes "users have problem Y" assumes "users currently do Z to cope"). Testing the most foundational assumption first avoids wasting effort validating downstream details on a broken foundation.

Characteristics of a Strong Hypothesis

• Specific — names the exact change, segment, and expected effect.
• Measurable — ties to a metric with a clear threshold for success/failure.
• Falsifiable — there's a real possibility the data proves it wrong.
• Time-bound — has a window in which the result will be evaluated.

A Practical Template

"We believe [this solution] will achieve [this outcome] for [this user segment]. We'll know we're right if we see [metric change] within [timeframe]."

Common Pitfalls

• Hypotheses so vague that any result can be claimed as a "win."
• Combining multiple variables in one hypothesis, making it impossible to know which one caused the result.
• Skipping the threshold — without "how much is enough," teams argue endlessly about ambiguous outcomes.

What is Assumption Mapping?

A visual technique to plot all assumptions behind a solution on two axes: importance (how critical to success) and evidence (how confident are we, based on current evidence).

The Assumption Map Grid

• High importance, low evidence — test these first; they're the riskiest unknowns.
• High importance, high evidence — proceed with confidence, but keep monitoring.
• Low importance — deprioritize testing regardless of evidence level; not worth the effort right now.

Using the Map in Practice

After brainstorming a solution, list every assumption it depends on (desirability, usability, feasibility, viability), plot them, and let the map decide your experiment roadmap — test top-right-quadrant assumptions before investing in build.

Connecting to Experimentation

Each high-importance, low-evidence assumption becomes a candidate hypothesis for the experiment design process (Module 10).