A test is how the team checks whether a hunch or assumption holds up. It names what's being checked, who it's checked with, how, and what result the team expects to see. Tests live in Testing, the third block in Measure, after Concepts and Hunches.
For example, the team thinks new cardholders ignore rewards because points feel abstract. To check, they plan a concept test with 10 new cardholders and expect 70% of them to succeed at the task if a new design makes point value clear. That plan is the test.
Why it matters
Without a written test, teams look at results and decide afterward what they mean. Almost any outcome can be read as good news, and hunches end up confirmed simply because someone wanted them to be.
Writing the test down before it runs fixes that. The expected result is set in advance, so everyone can see whether the evidence supports the hunch or goes against it. It also keeps tests focused: one hunch or assumption, the right people, and a clear measure.
Name what you're checking, agree on the expected result before you start, and pick an audience and method that can actually answer the question.
How we write one
Every test is written as one sentence that names the hunch, the audience, the method, and the expected result:
Test a hunch or assumption with an audience using a method; expect a result on a metric.
Test the abstract-points hunch with 10 new cardholders using a concept test; expect 70% task success on the point value task.
Writing it this way puts the expected result in the sentence itself, so the test can't be judged by moving the goalposts after the fact. Glare uses "test" in place of "hypothesis".
How it differs from similar ideas
-
A hunch or assumption is the belief being checked. A test is the plan for checking it.
-
A test question is one question inside a test. A test can hold several test questions.
-
A method is a named technique. In Testing it is the test type, such as a concept test; in Focus it is a comparison design, such as an A/B test, repeated across one variable. A test checks one hunch or assumption during Measure.
-
Findings are the bundle that comes out of testing. A test is the plan that goes in.
A good test
-
Checks one hunch or assumption
-
States the expected result before it runs
-
Names the audience and how many people
-
Uses a method that fits the question
-
Names one metric to judge the result
-
Has at least one test question
What happens to it
-
Planned. The team writes it down, or Ray finds it in a research plan or test brief.
-
Running. Sessions or data collection are under way, and observations start to come in.
-
Complete. Results are in and compared against the expected result.
-
Concluded. Signals are written that support or contradict the hunch, and they are bundled into Findings.
-
Cancelled. It was stopped before it finished; the reason is kept for history.
For Ray
This part tells Ray how to find a test in an input, such as a research plan, a test brief, a sprint board, a test readout or a planning email, and how to store it.
How to find one
-
Mark plans to check something. Read the input and mark every plan to find out whether a belief is true or a design works.
-
Find what's being checked. Link it to an existing hunch or assumption. If none exists, store the belief as a Hunch first.
-
Find the expected result. Pull out what the team expects to see if the belief is right. If it isn't stated, leave it blank and flag the test.
-
Find the audience and method. Identify who it's run with, how many, and the method used.
-
Find the metric. Identify the UX metric used to judge the result.
-
Find the test questions. Pull out each question the test must answer and store them as Test questions.
-
Write the sentence. Use the format: Test [hunch] with [audience] using [method]; expect [result] on [metric].
-
Fill the record, score it, link it and store it using the rules below.
Cues to look for
-
Plan language: "we'll test", "let's validate", "we want to find out whether", "our hypothesis is"
-
Research plans and test briefs listing participants, tasks and success criteria
-
Success thresholds: "we'd call it a win if 70%..."
-
Sprint tickets for usability sessions, concept tests or surveys
What doesn't count
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 50px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>If the statement...</p></th><th colspan="1" rowspan="1"><p>Store it as</p></th></tr><tr><td colspan="1" rowspan="1"><p>States a belief with no plan to check it</p></td><td colspan="1" rowspan="1"><p>Hunch or Assumption</p></td></tr><tr><td colspan="1" rowspan="1"><p>Is a single question to ask participants</p></td><td colspan="1" rowspan="1"><p>Test question</p></td></tr><tr><td colspan="1" rowspan="1"><p>Reports what happened in sessions</p></td><td colspan="1" rowspan="1"><p>Observation</p></td></tr><tr><td colspan="1" rowspan="1"><p>Repeats a measurement across one variable to compare options in Focus</p></td><td colspan="1" rowspan="1"><p>Method</p></td></tr><tr><td colspan="1" rowspan="1"><p>Summarizes the results of a round</p></td><td colspan="1" rowspan="1"><p>Findings</p></td></tr></tbody></table>
The record
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 75px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Field</p></th><th colspan="1" rowspan="1"><p>What to fill in</p></th><th colspan="1" rowspan="1"><p>Values</p></th></tr><tr><td colspan="1" rowspan="1"><p>Statement</p></td><td colspan="1" rowspan="1"><p>The one-sentence test</p></td><td colspan="1" rowspan="1"><p>Test [hunch] with [audience] using [method]; expect [result] on [metric].</p></td></tr><tr><td colspan="1" rowspan="1"><p>Checks</p></td><td colspan="1" rowspan="1"><p>The hunch or assumption being tested</p></td><td colspan="1" rowspan="1"><p>Link to Hunch or Assumption</p></td></tr><tr><td colspan="1" rowspan="1"><p>Expected result</p></td><td colspan="1" rowspan="1"><p>What the team expects if the belief holds</p></td><td colspan="1" rowspan="1"><p>Metric and threshold</p></td></tr><tr><td colspan="1" rowspan="1"><p>Audience</p></td><td colspan="1" rowspan="1"><p>Who it's run with, and how many</p></td><td colspan="1" rowspan="1"><p>Link to Audience, with count</p></td></tr><tr><td colspan="1" rowspan="1"><p>Method</p></td><td colspan="1" rowspan="1"><p>How it's run</p></td><td colspan="1" rowspan="1"><p>Concept test, usability test, survey, A/B test, or other</p></td></tr><tr><td colspan="1" rowspan="1"><p>Metric</p></td><td colspan="1" rowspan="1"><p>The measure used to judge it</p></td><td colspan="1" rowspan="1"><p>Link to UX Metric</p></td></tr><tr><td colspan="1" rowspan="1"><p>Test questions</p></td><td colspan="1" rowspan="1"><p>The questions it must answer</p></td><td colspan="1" rowspan="1"><p>Links to Test question</p></td></tr><tr><td colspan="1" rowspan="1"><p>Concept</p></td><td colspan="1" rowspan="1"><p>The concept it evaluates, if any</p></td><td colspan="1" rowspan="1"><p>Link to Concept, or blank</p></td></tr><tr><td colspan="1" rowspan="1"><p>Source</p></td><td colspan="1" rowspan="1"><p>The input and exact passage</p></td><td colspan="1" rowspan="1"><p>Link to Input</p></td></tr><tr><td colspan="1" rowspan="1"><p>Status</p></td><td colspan="1" rowspan="1"><p>Where it stands</p></td><td colspan="1" rowspan="1"><p>Planned, Running, Complete, Concluded, Cancelled</p></td></tr><tr><td colspan="1" rowspan="1"><p>Confidence</p></td><td colspan="1" rowspan="1"><p>How completely it was specified</p></td><td colspan="1" rowspan="1"><p>High, medium, low</p></td></tr></tbody></table>
New tests start as Planned. The expected result can't be changed once a test is Running; a new expected result means a new test.
Scoring confidence
Confidence measures how completely the test was specified before it ran, not whether it will succeed.
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 50px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Level</p></th><th colspan="1" rowspan="1"><p>When to use it</p></th></tr><tr><td colspan="1" rowspan="1"><p>High</p></td><td colspan="1" rowspan="1"><p>Written down before running, with a hunch, audience, method, metric and expected result</p></td></tr><tr><td colspan="1" rowspan="1"><p>Medium</p></td><td colspan="1" rowspan="1"><p>Planned in advance, but the expected result or metric is vague</p></td></tr><tr><td colspan="1" rowspan="1"><p>Low</p></td><td colspan="1" rowspan="1"><p>Pieced together after it ran, or the expected result is missing</p></td></tr></tbody></table>
If any part of the record can't be filled from the source, leave it blank and lower confidence by one level.
Linking
-
Check for duplicates first. If an existing test checks the same hunch with the same audience and method, add the new source to it instead of creating another.
-
Link to the hunch or assumption it checks and its source input.
-
Link to its test questions, audience and metric, and to the concept it shapes or evaluates.
-
Link the results. When observations, UX metrics and signals arrive, link them to the test and compare them to the expected result.
-
Note reruns. When a test is run again with changes, store it as a new test and link it to the earlier one.
When to store and when to ask
-
Store automatically when confidence is high.
-
Add to the review digest when confidence is medium or low, when the expected result is missing, or when the test checks a high-stakes hunch.
When to flag
-
A test has no expected result, so evidence can't be judged against it
-
A test checks nothing: no hunch or assumption is linked
-
The expected result was changed after the test started running
-
A test is Complete but no signals have been written for it after 14 days
-
A high-stakes hunch has no test planned
Worked example
Input: a test brief on the rewards team's sprint board.
"Concept test, week of Nov 3. Goal: find out if abstract points are why new cardholders ignore rewards. 10 participants, all cardholders under 90 days. Key task: say what 5,000 points buys. We'll call it a win at 70% task success with the new concept."
What Ray finds:
-
"Find out if abstract points are why new cardholders ignore rewards" points to the existing abstract-points hunch.
-
"10 participants, all cardholders under 90 days" is the audience, and "concept test" is the method.
-
"70% task success" is the expected result, and task success is the metric.
-
"Say what 5,000 points buys" becomes a separate Test question, linked to this test.
Stored record:
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 50px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Field</p></th><th colspan="1" rowspan="1"><p>Value</p></th></tr><tr><td colspan="1" rowspan="1"><p>Statement</p></td><td colspan="1" rowspan="1"><p>Test the abstract-points hunch with 10 new cardholders using a concept test; expect 70% task success on the point value task.</p></td></tr><tr><td colspan="1" rowspan="1"><p>Checks</p></td><td colspan="1" rowspan="1"><p>Hunch: We think people ignore rewards because points feel abstract.</p></td></tr><tr><td colspan="1" rowspan="1"><p>Expected result</p></td><td colspan="1" rowspan="1"><p>Task success at or above 70%</p></td></tr><tr><td colspan="1" rowspan="1"><p>Audience</p></td><td colspan="1" rowspan="1"><p>Participants: 10 cardholders under 90 days</p></td></tr><tr><td colspan="1" rowspan="1"><p>Method</p></td><td colspan="1" rowspan="1"><p>Concept test</p></td></tr><tr><td colspan="1" rowspan="1"><p>Metric</p></td><td colspan="1" rowspan="1"><p>Task success (Behavioral, Leading, Design KPI)</p></td></tr><tr><td colspan="1" rowspan="1"><p>Test questions</p></td><td colspan="1" rowspan="1"><p>Can participants say what 5,000 points buys?</p></td></tr><tr><td colspan="1" rowspan="1"><p>Concept</p></td><td colspan="1" rowspan="1"><p>Points progress bar</p></td></tr><tr><td colspan="1" rowspan="1"><p>Source</p></td><td colspan="1" rowspan="1"><p>Sprint board test brief, October 2026</p></td></tr><tr><td colspan="1" rowspan="1"><p>Status</p></td><td colspan="1" rowspan="1"><p>Planned</p></td></tr><tr><td colspan="1" rowspan="1"><p>Confidence</p></td><td colspan="1" rowspan="1"><p>High: written in advance with every part specified</p></td></tr></tbody></table>
Next step: stored automatically, and the abstract-points hunch moves from Raised to In test.
Edge cases
-
Called a hypothesis. "Our hypothesis is that..." is treated as a test if it includes a plan to check it, or as a Hunch if it doesn't.
-
No expected result. Store the test, leave the expected result blank, flag it, and prompt the team to set one before sessions begin.
-
One test, several hunches. If a test checks more than one hunch, link each one, with a separate expected result for each.
-
Mixed results. If some test questions pass and others fail, write a signal for each and let Findings carry the overall call.