Findings

Findings are the full bundle from one round of testing. They hold the concept that was tested, the hunches and test questions behind it, what was seen, what was measured, what it all means and what to do next.

For example, SecondBank tested a points progress bar with new cardholders. The Findings show that task success rose to 78% on the value test, and recommend moving the concept into iteration. Everything that led to that conclusion sits inside the same bundle.

Why it matters

Testing produces a lot of loose pieces: notes, numbers, quotes and opinions. When they're scattered across decks and chat threads, leaders get the headline without the support, and the same test gets rerun because nobody can find what it showed.

Findings keep a round of testing together. They give leaders one place to see the conclusion and the evidence behind it, they send the strongest signals on to the work that needs them, and they flag what to keep watching.

Close each round of testing with Findings that state the result, hold every supporting part and say what should happen next.

How we write one

Every set of Findings is summed up in one sentence that names the concept, the result, the test and the recommended next step:

A concept had a result on a test; recommend an action.

The progress bar raised task success to 78% on the value test; recommend moving to iteration.

Writing it this way leads with the concept and the result, ties them to the test that produced them and ends with what to do, so a leader can read the sentence and know where things stand. Findings open with the call on each hunch, supported or contradicted, and then give the design signal.

How it differs from similar ideas

  • A signal is one interpreted piece of evidence. Findings hold all the signals from a round of testing, along with everything else.

  • A recommendation is one suggested action. Findings carry the overall recommended next step and every individual recommendation.

  • A test is the plan for checking a hunch or assumption. Findings are what came back when the test was run.

  • A result is a realized business outcome measured against a business goal. Findings report on one round of testing, well before anything ships.

  • An input such as a research report is the raw document. Findings are the structured bundle Ray builds from it.

Good Findings

  • Name the concept and the test it was run against

  • State the result in one sentence

  • Hold all the observations and UX metrics from the round

  • Include signals that interpret the evidence

  • Give a clear recommended next step

  • Raise anything that needs watching

What happens to it

  1. Opened. Ray creates the bundle when a round of testing starts or when its first readout arrives.

  2. Filled. Observations, UX metrics, signals and recommendations are linked in as they're found.

  3. Concluded. The summary sentence is written and the recommended next step is set.

  4. Sent on. Signals go to initiatives, watch items are raised, and the concept is refined through component building.

  5. Archived. The bundle is kept as the record of that round, linked to the decisions it informed.


For Ray

This part tells Ray how to build Findings from an input, such as a research report, a test readout deck, a synthesis doc or a set of session notes, and how to store them. Findings are a container: Ray creates the bundle and links the parts, each stored by its own rules.

How to find one

  1. Identify the round of testing. Find the concept, the test and the dates. If a Findings bundle for this round already exists, add to it instead of opening a new one.

  2. Link the plan. Link the hunches, assumptions and test questions the round was meant to answer.

  3. Extract the evidence. Pull out every observation and UX metric, store each by its own rules, and link them in.

  4. Extract the meaning. Pull out every signal and recommendation, store each by its own rules, and link them in.

  5. Find the overall result. Identify the headline result on the test and the recommended next step. Open with the call on each hunch, supported or contradicted, then the design signal. If the input doesn't give one, leave it blank.

  6. Find what to watch. Note any metric or risk the input says to keep an eye on, and raise a Watch item for it.

  7. Write the sentence. Use the format: [Concept] [result] on [test]; recommend [action].

  8. Fill the record, score it, link it and store it using the rules below.

Cues to look for

  • Report structure: "summary", "key findings", "what we learned", "next steps"

  • A concept name paired with a test result: "the progress bar raised task success to..."

  • Readouts that close a round of sessions, sprints or concept tests

  • Headline recommendations: "recommend moving to iteration", "retest", "drop this direction"

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>Is one observation or metric from the round</p></td><td colspan="1" rowspan="1"><p>Observation or UX Metric, linked into the Findings</p></td></tr><tr><td colspan="1" rowspan="1"><p>Interprets one piece of evidence</p></td><td colspan="1" rowspan="1"><p>Signal, linked into the Findings</p></td></tr><tr><td colspan="1" rowspan="1"><p>Suggests one action</p></td><td colspan="1" rowspan="1"><p>Recommendation, linked into the Findings</p></td></tr><tr><td colspan="1" rowspan="1"><p>Reports business results after a change shipped</p></td><td colspan="1" rowspan="1"><p>Result or Business Outcome</p></td></tr><tr><td colspan="1" rowspan="1"><p>Describes the plan for a test that hasn't run yet</p></td><td colspan="1" rowspan="1"><p>Test</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 summary</p></td><td colspan="1" rowspan="1"><p>[Concept] [result] on [test]; recommend [action].</p></td></tr><tr><td colspan="1" rowspan="1"><p>Concept</p></td><td colspan="1" rowspan="1"><p>What was tested</p></td><td colspan="1" rowspan="1"><p>Link to Concept</p></td></tr><tr><td colspan="1" rowspan="1"><p>Test and test questions</p></td><td colspan="1" rowspan="1"><p>The plan the round answered</p></td><td colspan="1" rowspan="1"><p>Links to Test and Test question</p></td></tr><tr><td colspan="1" rowspan="1"><p>Hunches and assumptions</p></td><td colspan="1" rowspan="1"><p>What the round set out to check, with the call on each hunch: supported or contradicted</p></td><td colspan="1" rowspan="1"><p>Links to Hunch and Assumption</p></td></tr><tr><td colspan="1" rowspan="1"><p>Audience</p></td><td colspan="1" rowspan="1"><p>Who took part</p></td><td colspan="1" rowspan="1"><p>Link to Audience</p></td></tr><tr><td colspan="1" rowspan="1"><p>Observations and UX metrics</p></td><td colspan="1" rowspan="1"><p>The evidence collected</p></td><td colspan="1" rowspan="1"><p>Links to Observation and UX Metric</p></td></tr><tr><td colspan="1" rowspan="1"><p>Signals</p></td><td colspan="1" rowspan="1"><p>What the evidence means</p></td><td colspan="1" rowspan="1"><p>Links to Signal</p></td></tr><tr><td colspan="1" rowspan="1"><p>Recommendations</p></td><td colspan="1" rowspan="1"><p>Suggested next actions</p></td><td colspan="1" rowspan="1"><p>Links to Recommendation</p></td></tr><tr><td colspan="1" rowspan="1"><p>Recommended next step</p></td><td colspan="1" rowspan="1"><p>The overall call to make next</p></td><td colspan="1" rowspan="1"><p>Iterate, Retest, Move to Focus, Drop, other</p></td></tr><tr><td colspan="1" rowspan="1"><p>Watch items</p></td><td colspan="1" rowspan="1"><p>What to keep monitoring</p></td><td colspan="1" rowspan="1"><p>Links to Watch item</p></td></tr><tr><td colspan="1" rowspan="1"><p>Dates and source</p></td><td colspan="1" rowspan="1"><p>When the round ran and the input it came from</p></td><td colspan="1" rowspan="1"><p>Date range; 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>Open, Concluded, Sent, Archived</p></td></tr><tr><td colspan="1" rowspan="1"><p>Confidence</p></td><td colspan="1" rowspan="1"><p>How complete the bundle is</p></td><td colspan="1" rowspan="1"><p>High, medium, low</p></td></tr></tbody></table>

New Findings start as Open. They move to Concluded only when the summary sentence and recommended next step are filled.

Scoring confidence

Confidence measures how complete and well supported the bundle is, not whether the result was good.

<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>Concept, test, evidence, signals and a recommended next step are all present and linked</p></td></tr><tr><td colspan="1" rowspan="1"><p>Medium</p></td><td colspan="1" rowspan="1"><p>The result and evidence are present, but signals, recommendations or the test link are missing</p></td></tr><tr><td colspan="1" rowspan="1"><p>Low</p></td><td colspan="1" rowspan="1"><p>Only a headline result or partial notes, with little evidence linked</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. Each part linked into the bundle keeps its own confidence.

Linking

  • Check for duplicates first. If Findings already exist for the same concept, test and dates, add the new source and parts to them instead of creating another.

  • Link every part the round produced, and the source input.

  • Send signals to initiatives. Link each signal ready for Focus to the initiative it informs, or prompt for a new initiative if none fits.

  • Raise watch items and link them back to these Findings.

  • Link to the concept it refines, and to any earlier Findings on the same concept, so rounds can be compared.

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 its result contradicts an open test or earlier Findings, or when it recommends moving to a bet.

When to flag

  • Findings have no signals or no recommended next step

  • Findings were concluded but their signals were never sent to an initiative

  • A concept was moved forward with no Findings behind it

  • Two rounds of Findings on the same concept reach opposite results

  • Findings raised a watch item that has no owner

Worked example

Input: a research report on the SecondBank rewards value test, round 2.

"Summary: with the points progress bar, 78% of participants could say what 5,000 points buys, up from 42% on the current dashboard and above our 70% target. Participants still struggled to convert points, so we suggest showing dollar value. Recommend moving the progress bar to iteration. Keep an eye on redemptions once it ships."

What Ray finds:

  • "Points progress bar" is the concept, and the 5,000 points question ties it to the value test.

  • The 42% and 78% task success values are stored as UX metrics and linked in.

  • "Showing dollar value" is stored as a recommendation, and "moving to iteration" is the recommended next step.

  • "Keep an eye on redemptions" raises a Watch item linked to these Findings.

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>The progress bar raised task success to 78% on the value test; recommend moving to iteration.</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>Test and test questions</p></td><td colspan="1" rowspan="1"><p>Abstract-points test; "Can participants say what 5,000 points buys?"</p></td></tr><tr><td colspan="1" rowspan="1"><p>Hunches and assumptions</p></td><td colspan="1" rowspan="1"><p>Hunch: points feel abstract; Assumption: cardholders don't know rewards exist</p></td></tr><tr><td colspan="1" rowspan="1"><p>Audience</p></td><td colspan="1" rowspan="1"><p>Participants: new cardholders, first 90 days</p></td></tr><tr><td colspan="1" rowspan="1"><p>Observations and UX metrics</p></td><td colspan="1" rowspan="1"><p>7 of 10 scrolled past the balance; task success 42% and 78%</p></td></tr><tr><td colspan="1" rowspan="1"><p>Signals</p></td><td colspan="1" rowspan="1"><p>Point value is unclear; supports the abstract-points test</p></td></tr><tr><td colspan="1" rowspan="1"><p>Recommendations</p></td><td colspan="1" rowspan="1"><p>Show point value in dollars</p></td></tr><tr><td colspan="1" rowspan="1"><p>Recommended next step</p></td><td colspan="1" rowspan="1"><p>Iterate</p></td></tr><tr><td colspan="1" rowspan="1"><p>Watch items</p></td><td colspan="1" rowspan="1"><p>Watch redemptions</p></td></tr><tr><td colspan="1" rowspan="1"><p>Dates and source</p></td><td colspan="1" rowspan="1"><p>October 2026; research report, round 2</p></td></tr><tr><td colspan="1" rowspan="1"><p>Status</p></td><td colspan="1" rowspan="1"><p>Concluded</p></td></tr><tr><td colspan="1" rowspan="1"><p>Confidence</p></td><td colspan="1" rowspan="1"><p>High: all parts present and linked</p></td></tr></tbody></table>

Next step: stored automatically; the signal is sent to the rewards visibility Initiative and the redemptions Watch item is raised.

Edge cases

  • Findings spread across several inputs. A readout deck, notes and an email may all cover one round. Build one bundle and link every input as a source.

  • No clear result. If a round ends without a headline, conclude the Findings with the result left blank and the next step set to Retest, at medium confidence.

  • Results that disagree. If metrics improved but signals point the other way, keep both in the bundle and send it to the review digest.

  • Several concepts in one round. A test that compares two concepts produces one bundle per concept, linked to each other for comparison.

Identify where decision quality breaks down

The Glare Design Assessment helps teams spot weak validation, stakeholder friction, alignment gaps, and assumptions that scale without measurable learning—so you have a clearer starting point for improvement.

About 5 minutes · Team-based · Diagnostic snapshot you can act on

Take the Design Assessment
Findings | Glare