User Research Report
beta · Family research-docs · Phase discover · Sizes lean, full · ~4,650 tokens
The document written after one study of the people a product serves: what the study found, how it answers the questions the plan set out to ask, and what to do about it. For a usability test it also carries the tasks’ results against the measures the plan set, which is why practitioners sometimes call it a usability test report.
This document records what one study of the people a product serves found, and what it recommends,
after the study ends. It is not the user research plan, the sibling bundle that sets the same
study’s questions before the sessions run; this report answers them after. The deep explainer behind
every rule here is user-research-report_companion.md; a fully
worked instance is user-research-report_example.md.
Before you start: is this the right document at all?
Section titled “Before you start: is this the right document at all?”Write a user research report once a study has run and you need to tell stakeholders what it found, what to do about it, and whether it answered the questions the plan set out to ask. Dovetail separates the report from its parts: “A report packages several insights into a narrative for a specific audience, with method, context, and recommendations.” User Interviews names the same shape: “A research report is a document that summarizes all the details of a research study, including the research questions, methodology, notable insights, and recommended next steps.”
Write something else if:
| You actually need | Because |
|---|---|
| a user research plan | the not-yet-run sibling to this bundle; it sets out the same study’s questions before the sessions happen, and this report answers them after |
| a debrief | Dovetail’s own line draws the boundary: “A debrief is not a research report for peer review.” A debrief is a lighter, often async record a team can lean on quickly; it does not replace this document, though “the async debrief still serves as the foundation” for it |
a test-summary-report (the QA pair’s own report, qa-docs) |
that document verifies a product increment against an agreed specification and grades it; this report studies how people use a product and records observations, not a verdict against a spec |
an experiment readout (experimentation-docs) |
Gilad draws the line this library relies on: “Per this definition an A/B test is an experiment, but a usability test is not.” A usability study assigns no participant to a variant and applies no decision rule against a baseline |
a persona, a problem statement, or a design brief (discovery-docs) |
this family’s own contract states the boundary directly: “The persona, the problem statement and the design brief take the findings and frame or decide.” This report may recommend; it frames no problem and chooses no concept |
| a Voice of Customer report, or any standing research repository | those aggregate findings across many studies, ongoing; this report is written for, and finishes with, the one study it reports |
Avoid the bare phrase “test report” or “test results” for this document. The words collide with
test-summary-report, a separate bundle in this library’s QA family. Name this one fully, as a
usability study report or a user research report.
Who writes it. No source read for this bundle gives one role a majority. GitLab names the role, not the credential: “In GitLab, anyone can be the DRI of a UX Research project. If the Research DRI is not a UX Researcher, they are partnered with a UX Researcher acting in a supporting role.” User Interviews assumes the researcher who ran the study writes it up. This library lets either a researcher or a product person write this report, without claiming either one usually does.
A report with no written plan behind it still fills Research Questions and Answers. The sources tie a report to a plan only where one exists; if no plan was written, fill that section from what the study actually set out to learn.
Picking a variant
Section titled “Picking a variant”Lean carries the six sections every size ships, in this order: Summary, Research Questions and Answers, What We Did, Key Findings, Task Results, and Recommendations. Full adds three, in this order: Background and Objectives (after Summary), Open Questions (after Recommendations), and Appendix (after Open Questions).
Use full when at least one of these is true: the decision these findings inform needs restating with its own context for a reader who never saw the plan; a question the study raised and could not answer needs its own place, rather than folding into a single status column; or the session list, the materials used, and where the raw notes live are worth pointing to in their own section. Use lean when none of those is true: the decision is already clear from Summary, every question the plan set gets a clean status in Research Questions and Answers, and nobody downstream needs the raw materials linked from this document itself.
Task Results ships at every size, filled in only for a usability test. For any other method - interviews, a diary study, a survey, a field visit - the section is one line naming the method actually used, with the task table and the questionnaire line marked N/A. That is a complete answer, not a placeholder waiting to be filled in later. The content this block carries, task completion against a stated criterion, errors, time, and questionnaire scores, comes from usability-specific sources only; no general report list in this bundle’s research carries it, which is why it is conditional rather than dropped at the smaller size.
Two depths, not three, at medium confidence. This is where the report differs from the plan: no source read for the plan’s own bundle names a short and a long form of one plan, but several name a short and a long form of one report. Dovetail states the two audiences directly: “Executives want one page with the decision; your product team wants the detail behind it.” UserTesting names a third, shortest form too, a “Quick feedback snapshot,” but its four listed items each map onto a lean section this bundle already ships, which is why this library treats it as a restatement of lean rather than a third size.
The rubric
Section titled “The rubric”Below 13 out of 18 on a full report, a reader cannot tell what the study found, what to do about it, or how far a thin sample lets them trust it, without going back to ask the author questions this document should have answered on its own. Below 8 out of 12 on a lean report, the same gap opens with three fewer sections standing between a weak answer and the stakeholder reading it.
| # | Criterion | 0 | 1 | 2 |
|---|---|---|---|---|
| 1 | Answer comes first | Summary opens with the method, the participant count, or a vague statement, with no finding stated | A finding is stated, but the recommendation is missing, or only restates the finding rather than naming an action | A reader who stops after Summary already knows what was found, what the team should do about it, and, where it is named, what decision that informs and who decides it |
| 2 | Each question answered | The section summarizes themes across the study instead of going question by question, or a question is answered more broadly than the plan asked it | Each plan question gets its own row, but the hypothesis column (confirmed, contradicted, or N/A) or the Status column (answered or open) is missing or inconsistent for at least one | Every question the plan set gets its own answer, on the plan’s own terms, backed by a named finding, with a status a reader can check against Key Findings |
| 3 | Method and limits named | Participants are reported with no denominator, so a reader cannot tell how many actually ran against how many were recruited, or no limit is stated for what this sample lets you claim | A fraction with its denominator is given, but the stated limit is generic rather than specific to what this sample size can and cannot support | The method is named and matches the plan, or a deviation is named and why; participants are a fraction with a denominator; and the stated limit names specifically what this sample lets a reader conclude and what it does not |
| 4 | Findings carry checkable evidence | A finding is an opinion with a number attached, no quote or observed action a reader could go check, and no positive finding appears anywhere | Evidence exists, but a count is given as a bare percentage with no denominator stated nearby, or a participant’s own suggestion is folded into the team’s recommendation instead of labeled as the participant’s words | Every finding names a pattern, a fraction with its denominator, and a quote or observed action a reader could check; at least one positive finding carries the same evidence standard as the problems; and a participant’s own suggestion is labeled as such |
| 5 | Task Results scoped correctly | A usability test’s table has no stated success criterion behind its completion numbers, or a study that is not a usability test leaves the table blank with no line explaining why | A criterion or an N/A line is present, but thin: a completion rate with the criterion unstated, or an N/A line that names no actual method | For a usability test, every task states completion as a fraction against the plan’s own success criterion, with errors and time where the plan collected them; for any other method, one line names the method used and marks the table and the questionnaire line N/A |
| 6 | Recommends without deciding | A recommendation names a destination with no finding behind it, or the document states what the team has decided, or names the concept it will build | Every recommendation ties to a finding, but at least one does not say why that finding makes the change worth weighing, or the line between a recommendation and a decision blurs somewhere in the section | Every recommendation names a specific change, points at the finding or findings it answers to, states why that finding makes the change worth weighing, and nowhere does the document name the decision made or the concept chosen |
| 7 | Background links the plan (full only) | The section restates the study’s title, with no link to the plan and no decision restated | The plan is linked and the decision is restated, but, where this studies something people already use, the reasoning for why that is still discovery work is missing or asserted with no reasoning attached | A reader who has not seen the plan gets why the study ran, the decision restated, a working link to the plan, and, where the product already ships, a stated reason the study still asks the generative question rather than grading a spec |
| 8 | Open Questions point forward (full only) | A question restates something the study already answered, just more vaguely, or names a vague area of curiosity with no path to an answer | The question is specific, but the section does not say why this study could not answer it, or does not say what a future study would need to do differently | Each open question is specific and checkable, names why this study could not answer it, and names what a future study would need to do differently to answer it |
| 9 | Appendix points to policy (full only) | No session list or materials a reader could actually open, or a retention period stated as a figure this study set | The session list and materials are linked, but the raw notes’ location names no retention policy, or restates a figure that belongs to a standing program as though it were this study’s own | The session list, the materials used, and where the raw notes live are all things a reader could open, and retention points to the organization’s own standing policy rather than a figure this report restates as its own |
Which rows apply to what. Full ships all nine rows, because a reader of a full report can ask about the restated background, the questions the study genuinely could not reach, and where the raw materials live, as well as the six every size must get right. Lean ships the first six, every row the lean variant has a section to support.
| Document | Rows | Maximum | Score against |
|---|---|---|---|
| lean | 1-6 | 12 | 8 |
| full | all 9 | 18 | 13 |
The test behind every row above: could someone satisfy the cell by adding words without the report actually telling a reader more? Row 4’s count, for instance, is never graded by how many findings are listed; it is graded by whether a reader could go check the evidence behind any one of them, which is the one thing padding cannot fake.
Named anti-patterns
Section titled “Named anti-patterns”Opening with the method or the participant count instead of the answer. The tell: a reader has to get past Background or What We Did before learning what the study found. Fix by stating what was found and what to do about it in Summary’s first lines, the way Dovetail puts it: “Put the headline finding in the first three sentences-don’t make readers excavate it.”
Answering a broader question than the one the plan asked. The tell: Research Questions and Answers reads like a theme summary rather than a row-by-row answer to the plan’s own list. Fix by taking each question in the plan’s own words, giving it a short answer backed by a named finding, and marking plainly whether it stayed open.
Reporting a participant count with no denominator. The tell: “9 participants,” with no stated number recruited, so a reader cannot tell how many actually ran against how many were asked. Fix by stating participants as a fraction, the way Sauro and Lewis recommend: “Unless this information appears elsewhere in your report, we recommend presenting fractions.”
Reporting an opinion with a number attached instead of a finding with evidence. The tell: a row in Key Findings names a feeling, such as “users don’t like X,” with a count and no quote or observed action behind it. Fix by naming the pattern, the fraction, and a quote or action a reader could go check. Dovetail states the failure directly: “A finding without evidence is an opinion with formatting.”
Reporting a completion rate with no stated success criterion. The tell: Task Results gives a percentage with no sentence saying what counted as success. Fix by stating the plan’s own criterion beside every completion fraction, the way NIST defines the measure itself: “The total number of successes were calculated for each task and then divided by the total number of times that task was attempted.”
Naming the concept the team should build, or stating that a decision has been made. The tell:
Recommendations reads like a decision record instead of a list of changes worth weighing. Fix by
naming what is worth weighing and why, and leaving the decision itself, and the chosen concept, to a
later document such as design-brief or problem-statement.
Restating an organization’s retention period as though it were this study’s own figure. The tell: a number of months in Appendix that reads copied from somewhere else rather than set by this study. Fix by pointing to the organization’s standing policy instead, as GOV.UK instructs: “Keep your research data only in places that your organisation has approved for storing personal data.”
Calling this document a bare “test report” or “test results.” The tell: the word “test” appears
with no “research,” “usability,” or “user” attached, inviting confusion with test-summary-report, the
QA pair’s own document. Fix by naming the document fully, a usability study report or a user research
report, never a bare “test report” or “test results.”
When it is good enough
Section titled “When it is good enough”When a reader who stops after Summary already knows what was found and what to do about it, when every question the plan set gets its own checkable answer, when every finding carries evidence a reader could go verify rather than an opinion with a number attached, and when the document recommends without quietly deciding, the report has done its job.
Then self-grade against the rubric above, and delete every HTML comment. A report that reads well but leaves a finding’s evidence to the reader’s trust, or that slides from recommending into deciding, has not finished the job it exists to do, whatever else it contains.
The artifacts
Section titled “The artifacts”user-research-report_template-lean.md · ~4,650 tokens
---title: "{{study_name}} Research Report"study_name: "{{study_name}}"author: "{{author}}"plan_link: "{{plan_link}}"status: "{{status}}"last_updated: "{{date}}"doc_type: user-research-reportsize: leansource_template: user-research-reportsource_template_version: 0.1.0---
<!--LEAN USER RESEARCH REPORT. The shortest form that still answers the plan's own questions, states whatthe study found, and recommends what to do next, with Task Results carried at every size because thecatalog alias "usability test report" promises a reader it is there. Use it for one audience whoneeds the decision and the evidence behind it, not the study's full background or its raw materials.To add Background and Objectives, Open Questions, and Appendix, seeuser-research-report_template-full.md. Never rename or reorder the sections below; the full variant isa strict ordered superset of this one.
WHAT A USER RESEARCH REPORT IS, AND IS NOTIt is written after one study of the people a product serves, and it answers the questions a plan setbefore the sessions ran. Dovetail: "A report packages several insights into a narrative for a specificaudience, with method, context, and recommendations." User Interviews: "A research report is adocument that summarizes all the details of a research study, including the research questions,methodology, notable insights, and recommended next steps." IT IS NOT the plan: the plan sets thequestions before the study runs; this report answers them after. IT IS NOT a decision record: thisreport recommends, and a recommendation names a change worth weighing, but it names no concept aschosen and records no decision. That is a later document's job, such as `design-brief` or`problem-statement`. IT IS NOT a QA test report: `test-summary-report` verifies a product incrementagainst an agreed specification and grades it; this report studies how people use a product andrecords observations, not a verdict against a specification. Avoid the bare phrase "test report" or"test results" for either document; name this one fully, as a usability study report or a userresearch report. IT IS NOT an experiment readout: a usability study assigns no participant to avariant and applies no decision rule against a baseline, which is the line this library draws betweena test and an experiment. IT STILL DECLARES DISCOVERY WORK FOR A SHIPPED PRODUCT: a study that askswhere people lose time with something they already have is still asking the generative question, notthe evaluative one, whatever the product's own age. See user-research-report_companion.md sections 1,6, 7 and 8.
HOW TO FILL THIS IN1. Read the comment under each heading: WHAT it wants, WHY it matters (with a pointer into user-research-report_companion.md), guiding questions to ASK, a GOOD and a WEAK example, and the TRAP to avoid. For a table, PRIORITY explains the ordering rule and ROW HINT says what a good row contains.2. Replace each {{placeholder}} with your content.3. If a section does not apply, including Task Results when this study is not a usability test, write "N/A" and one line of why, rather than deleting the section.4. Before you share this report: self-grade against user-research-report_guide.md, then DELETE every HTML comment. They are guidance, not content.-->
# {{study_name}} Research Report
## Summary
<!-- WHAT The answer up front: what the study found and what it recommends, in a few lines. May also name the decision these findings inform and who makes it. WHY Dovetail: "Put the headline finding in the first three sentences-don't make readers excavate it". Maze: "Lead with a short summary of key findings and recommendations, then add the context people need-such as why you conducted the study and the methods you used." NIST's template, customized for usability testing of Electronic Health Records (EHRs), opens the same way: "1 EXECUTIVE SUMMARY". No shipped source names this pairing as its own heading, but Dovetail's "End each finding with what the team should do about it, and who decides" and Useberry's "Lead with decisions, then show the evidence" both put the decision at or near the top, not buried in a methodology section. Deep dive: user-research-report_companion.md section 3 (Anatomy > Summary). ASK What did the study find, and what does it recommend, in a few lines? What decision do these findings inform, and who makes it, if naming that helps the reader? GOOD "Found: people delete duplicate photos one at a time because the bulk-select control reads as a filter toggle; 7 of 9 participants never used it. Recommend: make bulk-select the default view once duplicates are found. Decision this informs: whether to prioritize the redesign next sprint, decided by the mobile lead." WEAK "Participants had some issues with duplicate photos and we have some ideas." (no answer stated, no recommendation, no decision named) TRAP Opening with the method or the participant count instead of the answer. A reader who stops after one paragraph should already know what the study found. -->
- What the study found, and what it recommends: {{headline_summary}}- Decision these findings inform, and who makes it: {{decision_and_owner}}
## Research Questions and Answers
<!-- WHAT Each question the plan set, with a short answer and the findings behind it. Whether the answer confirms the plan's hypothesis, where it had one. A question the study could not answer says so. WHY UserTesting's detailed report form carries "Research Questions" as its own item. User Interviews ties the section to the plan directly: "Actionable reports should relate back to the initial research question: How do your findings influence the decisions you mapped out in your research plan?" This library holds, as its own position drawn from this family's contract, that a report answers the questions the plan set out. Anderson-Stanier names the failure mode this row guards against: "There are questions out there that qualitative, and even quantitative, research simply cannot answer and we need to let our teams know when a question is unanswerable." Deep dive: user-research-report_companion.md section 3 (Anatomy > Research Questions and Answers). ASK What was the question, in the plan's own words? What is the short answer, and which findings support it? If the plan stated a hypothesis, did the answer confirm or contradict it? Did the question stay open, and if so, why? PRIORITY One row per question the plan set, in the plan's own order. Add no new questions here; a pattern the study turned up that the plan never asked belongs in Key Findings instead. ROW HINT A good row restates the question, gives a short answer backed by a finding, and says plainly whether it confirmed a stated hypothesis or stayed open. A weak row answers a different, broader question than the one asked. GOOD | What makes someone miss the bulk-select control? | It reads as a filter toggle; 7 of 9 missed it on first use. | Confirmed the plan's hypothesis. | Answered | WEAK | How do people feel about duplicate photos? | They find them annoying. | | Answered | TRAP Summarizing the study's themes instead of answering the plan's own questions one by one, including a question nobody happened to raise in conversation but that the plan still asked. -->
| Research question (as the plan wrote it) | Answer | Confirms the plan's hypothesis? | Status ||---|---|---|---|| {{research_question}} | {{research_answer}} | {{confirms_hypothesis}} | {{question_status}} |
## What We Did
<!-- WHAT The method, the participants, and the dates, the limits of the sample stated as a fraction with its denominator, and one line on anything that ran differently from the plan. WHY Dovetail's own list carries "method and participants". NIST's template, inside the same EHR customization, structures the same content under "3 METHOD" and "3.1 PARTICIPANTS". UserTesting's detailed form carries "Methodology (e.g., unmoderated usability tests, surveys, etc.)". Sauro names the specific limit a small sample carries: "the key limitation is that you are limited to detecting large differences between designs or measures". The one-line deviation note is this library's own arrangement, the same device `experiment-readout` carries in its own Context and Setup section; no source instructs a report to note a deviation from its own plan. Deep dive: user-research-report_companion.md section 3 (Anatomy > What We Did). ASK What method did the study use, and does it match the plan? Who took part, stated as a fraction of those recruited? When did sessions run, and which version or state of the product were participants reacting to? What does a sample this size let you claim, and what does it not let you claim? Did anything run differently from the plan, and if so, what and why, in one line? GOOD "One-to-one moderated interviews with a walkthrough of each participant's own camera roll, 2026-03-10 to 2026-03-14. 9 of 10 recruited completed a session; one dropped for a scheduling conflict. A sample this size can surface a large, frequent problem; it cannot rank two close icon designs against each other. Ran as planned, except two sessions moved to phone calls when video access failed." WEAK "We talked to some users and took notes." (no method named, no participant count or denominator, no dates, no stated limit) TRAP Reporting a participant count with no denominator, so a reader cannot tell how many actually ran against how many were recruited. -->
- Method, and whether it matches the plan: {{method_used}}- Participants, as a fraction of those recruited: {{participant_summary}}- Dates sessions ran: {{session_dates}}- What this sample size lets you claim, and what it does not: {{sample_limits}}- One line on anything that ran differently from the plan: {{plan_deviation_note}}
## Key Findings
<!-- WHAT Each finding, stated as a pattern rather than a single anecdote, how many participants showed it as a fraction with its denominator, and the evidence, a quote or an observed action, a reader can check. Positive findings alongside problems, and a participant's own suggestion labelled as the participant's words. WHY NN/g names the chain this section sits at the end of: "Data leads to findings, which build insights, which justify recommendations." Dovetail: "A one-page summary with linked evidence beats a forty-slide deck nobody opens." Dovetail, more bluntly: "A finding without evidence is an opinion with formatting." Sauro and Lewis prefer fractions over a bare percentage: "Unless this information appears elsewhere in your report, we recommend presenting fractions." Schade: "Don’t forget to include positive findings as well, letting the team know what’s already working." GOV.UK: "Whenever you can, you should fully anonymise the extracts you use so that the participants cannot be identified." User Interviews names the suggestion row directly: "Any specific recommendations brought up by users during interviews or in surveys." Deep dive: user-research-report_companion.md section 3 (Anatomy > Key Findings). ASK What is the finding, as a pattern across sessions? How many participants showed it, as a fraction with the denominator stated? What evidence backs it, anonymised and within the consent given? Is this a problem, something that already works, or a participant's own suggestion? Have you included at least one thing that already works, not only what failed? PRIORITY Most important finding first. State counts as fractions with the denominator next to every count, never as a bare percentage unless the denominator already appears beside it. ROW HINT A good row names a pattern, gives a fraction with its denominator, and quotes or describes evidence a reader could go check. A weak row is an opinion with a number attached and no quote or observation behind it. GOOD | Bulk-select reads as a filter toggle, so people delete duplicates one at a time | 7 of 9 | "I thought that button was for sorting, not for picking more than one." | Problem | WEAK | Users don't like managing duplicates | Most of them | | Problem | TRAP Reporting an opinion with a number attached instead of a pattern with a quote or an observed action a reader could go check. -->
| Finding | Participants (fraction) | Evidence | Positive, problem, or participant suggestion ||---|---|---|---|| {{finding}} | {{finding_count}} | {{finding_evidence}} | {{finding_type}} |
## Task Results
<!-- WHAT For a usability test: each task's completion, errors and time against the measures the plan set, and the questionnaire scores the plan named. For any other method, one line saying so, with the table and the questionnaire line marked N/A. WHY This block's content comes from usability-specific sources only; no general report list carries it. NIST's template, inside its own health-records customization, structures it under "3.9 USABILITY METRICS" and "4 RESULTS", and defines a success rate directly: "The total number of successes were calculated for each task and then divided by the total number of times that task was attempted." UserTesting instructs a report to "Include quantitative data where applicable (e.g., success rates, charts, QX Score visualizations)". Questionnaire thresholds, where NIST's template states them inside the same EHR customization, are NIST's own convention, not a general usability bar this library sets. Deep dive: user-research-report_companion.md section 3 (Anatomy > Task Results). ASK Is this a usability test? If not, say so in one line and mark the table and the questionnaire line N/A. If so: did each participant complete each task, against the success criterion the plan set? How many errors, and how long did it take? What did any questionnaire score, and against what convention, if one applies? PRIORITY List tasks in the plan's own order. A success rate is the count that actually succeeded divided by the count that attempted the task, never a rate scored after the fact against a looser bar. ROW HINT A good row names the task, states completion as a fraction against the plan's success criterion, gives error and time figures where the plan collected them. A weak row gives a completion rate with no stated criterion behind it. GOOD | Find and merge the duplicate pair in this album | 8 of 9 | 1 minor | avg 42s | Plan's success criterion: merges without opening the help menu | WEAK | Duplicate task | 90% | | | | TRAP Reporting a completion rate with no criterion stated, so a reader cannot tell what counted as success. -->
- Method, if this study is not a usability test (the table and the questionnaire line are then N/A): {{not_usability_test_note}}
| Task | Completion (fraction, against the plan's criterion) | Errors | Time on task | Notes ||---|---|---|---|---|| {{task_name}} | {{task_completion}} | {{task_errors}} | {{task_time}} | {{task_notes}} |
- Questionnaire scores the plan named, and the convention they are read against, if any: {{questionnaire_scores}}
## Recommendations
<!-- WHAT What the team should do next, and why, with each recommendation tied to one or more findings above. It records no decision and names no chosen concept. WHY Dovetail's list carries "recommendations". User Interviews names "Recommendations" as its own section, and UserTesting prioritises it: "Prioritized based on impact". Schade's own instruction keeps the section honest about what it is: "But present recommendations as recommendations." The sources pull this section toward the decision itself: Dovetail: "End each finding with what the team should do about it, and who decides". Usercall: "If your report doesn't end in a decision or a debate about a decision, you haven't written a report, you've written a diary." Against that pull, this family's contract states the line this bundle follows: "A member here may recommend, but it records no decision." Deep dive: user-research-report_companion.md section 3 (Anatomy > Recommendations), section 6. ASK What should the team do next? Which finding or findings does this recommendation answer to? Why this change, rather than another? Have you stopped short of naming the decision itself, or the concept a team will build? PRIORITY Highest-impact recommendation first. Every recommendation names the finding it answers to; a recommendation with no finding behind it is an opinion, not a conclusion. ROW HINT A good row names a specific change, points at the finding it comes from, and says why that finding makes the change worth weighing. A weak row names a destination ("fix duplicates") with no finding and no reasoning behind it. GOOD | Make bulk-select the default view once duplicates are found | Bulk-select reads as a filter toggle (7 of 9) | Removes the step where people have to notice a control they currently miss | WEAK | Redesign the duplicates feature | | Seems like a good idea | TRAP Naming the concept the team should build, or stating that the team has decided to build it. That is a later document's job; this section names what is worth weighing and why. -->
| Recommendation | Tied to finding(s) | Why ||---|---|---|| {{recommendation}} | {{tied_finding}} | {{recommendation_rationale}} |user-research-report_template-full.md · ~6,600 tokens
---title: "{{study_name}} Research Report"study_name: "{{study_name}}"author: "{{author}}"plan_link: "{{plan_link}}"status: "{{status}}"last_updated: "{{date}}"doc_type: user-research-reportsize: fullsource_template: user-research-reportsource_template_version: 0.1.0---
<!--FULL USER RESEARCH REPORT. Everything the lean variant carries, plus Background and Objectives (afterSummary), Open Questions (after Recommendations), and Appendix (after Open Questions).
THIS VARIANT IS A STRICT ORDERED SUPERSET OF THE LEAN ONE. The six lean sections appear here under thesame headings, in the same order, with the same placeholders; three sections are inserted among them.If you started lean, grow into this without rewriting anything you already filled in.
WHY THREE MORE SECTIONS, AND WHY THE SPLIT IS THIS LIBRARY'S OWN. This bundle ships two depths, notthree, at medium confidence. Dovetail states the two audiences directly: "Executives want one pagewith the decision; your product team wants the detail behind it." UserTesting names them as namedproducts, an "Executive summary report (For Stakeholders and Leadership)" and a "Detailed researchreport (For UX and Product teams)". UserTesting also names a third, shortest form, a "Quick feedbacksnapshot (For Agile teams)". Each of its four listed items maps onto a lean section this bundlealready ships, so this library treats that snapshot as a restatement of lean, not a third size. Whattips a report from lean to full: a study that needs its decision restated with its own background,the questions it could not answer named in their own section rather than folded into ResearchQuestions and Answers, and the raw materials pointed to in an appendix. Seeuser-research-report_companion.md section 4.
WHAT A USER RESEARCH REPORT IS, AND IS NOTIt is written after one study of the people a product serves, and it answers the questions a plan setbefore the sessions ran. Dovetail: "A report packages several insights into a narrative for a specificaudience, with method, context, and recommendations." User Interviews: "A research report is adocument that summarizes all the details of a research study, including the research questions,methodology, notable insights, and recommended next steps." IT IS NOT the plan: the plan sets thequestions before the study runs; this report answers them after. IT IS NOT a decision record: thisreport recommends, and a recommendation names a change worth weighing, but it names no concept aschosen and records no decision. That is a later document's job, such as `design-brief` or`problem-statement`. IT IS NOT a QA test report: `test-summary-report` verifies a product incrementagainst an agreed specification and grades it; this report studies how people use a product andrecords observations, not a verdict against a specification. Avoid the bare phrase "test report" or"test results" for either document; name this one fully, as a usability study report or a userresearch report. IT IS NOT an experiment readout: a usability study assigns no participant to avariant and applies no decision rule against a baseline, which is the line this library draws betweena test and an experiment. IT STILL DECLARES DISCOVERY WORK FOR A SHIPPED PRODUCT: a study that askswhere people lose time with something they already have is still asking the generative question, notthe evaluative one, whatever the product's own age. See user-research-report_companion.md sections 1,6, 7 and 8.
HOW TO FILL THIS IN1. Read the comment under each heading: WHAT it wants, WHY it matters (with a pointer into user-research-report_companion.md), guiding questions to ASK, a GOOD and a WEAK example, and the TRAP to avoid. For a table, PRIORITY explains the ordering rule and ROW HINT says what a good row contains.2. Replace each {{placeholder}} with your content.3. If a section does not apply, including Task Results when this study is not a usability test, write "N/A" and one line of why, rather than deleting the section.4. Before you share this report: self-grade against user-research-report_guide.md, then DELETE every HTML comment. They are guidance, not content.-->
# {{study_name}} Research Report
## Summary
<!-- WHAT The answer up front: what the study found and what it recommends, in a few lines. May also name the decision these findings inform and who makes it. WHY Dovetail: "Put the headline finding in the first three sentences-don't make readers excavate it". Maze: "Lead with a short summary of key findings and recommendations, then add the context people need-such as why you conducted the study and the methods you used." NIST's template, customized for usability testing of Electronic Health Records (EHRs), opens the same way: "1 EXECUTIVE SUMMARY". No shipped source names this pairing as its own heading, but Dovetail's "End each finding with what the team should do about it, and who decides" and Useberry's "Lead with decisions, then show the evidence" both put the decision at or near the top, not buried in a methodology section. Deep dive: user-research-report_companion.md section 3 (Anatomy > Summary). ASK What did the study find, and what does it recommend, in a few lines? What decision do these findings inform, and who makes it, if naming that helps the reader? GOOD "Found: people delete duplicate photos one at a time because the bulk-select control reads as a filter toggle; 7 of 9 participants never used it. Recommend: make bulk-select the default view once duplicates are found. Decision this informs: whether to prioritize the redesign next sprint, decided by the mobile lead." WEAK "Participants had some issues with duplicate photos and we have some ideas." (no answer stated, no recommendation, no decision named) TRAP Opening with the method or the participant count instead of the answer. A reader who stops after one paragraph should already know what the study found. -->
- What the study found, and what it recommends: {{headline_summary}}- Decision these findings inform, and who makes it: {{decision_and_owner}}
## Background and Objectives
<!-- WHAT Why the study ran, the decision it informs restated with more context than Summary gives, a link to the plan, and, if this studies a product people already use, why that is still discovery work. WHY Dovetail's own list carries "background and objectives (two or three sentences on why the study ran)". UserTesting names "Background & Objectives" in its detailed form, and the same idea under a different name in its shorter forms: "Objective of the Study" and "Study Goal". Maze's own instruction is to keep this brief: "Your introduction should lay out your research goals, plan, and scope." Every short form that names an objective at all puts it near the top of the document, which is why Summary carries it briefly at every size and this section expands it in full, with a link back to the plan. Deep dive: user-research-report_companion.md section 3 (Anatomy > Background and Objectives), section 2. ASK Why did this study run, with more detail than Summary gave? What decision does it inform, restated here for a reader who has not seen the plan? Where does the plan live? If this studies something people already use, why is it still discovery work rather than a check against a specification? GOOD "This study ran because support tickets mentioning duplicate photos have risen for two quarters with no clear cause. Decision it informs: whether to fund a redesign of the duplicate-cleanup flow next quarter. Plan: linked below. Though people already use this feature, the study asks where they lose time with it today, not whether it meets a spec, so it is still discovery work." WEAK "We wanted to learn more about duplicate photos." (no decision restated, no link to the plan, no reasoning on why a study of a shipped feature is still discovery work) TRAP Treating this section as a restatement of the study's title instead of linking to the plan and restating, for a reader who has not seen it, the decision the findings inform. -->
- Why this study ran: {{why_study_ran}}- Decision it informs, restated for a reader who has not seen the plan: {{decision_informed_detail}}- Link to the plan: {{plan_link_detail}}- If this studies something people already use, why it is still discovery work: {{shipped_product_note}}
## Research Questions and Answers
<!-- WHAT Each question the plan set, with a short answer and the findings behind it. Whether the answer confirms the plan's hypothesis, where it had one. A question the study could not answer says so. WHY UserTesting's detailed report form carries "Research Questions" as its own item. User Interviews ties the section to the plan directly: "Actionable reports should relate back to the initial research question: How do your findings influence the decisions you mapped out in your research plan?" This library holds, as its own position drawn from this family's contract, that a report answers the questions the plan set out. Anderson-Stanier names the failure mode this row guards against: "There are questions out there that qualitative, and even quantitative, research simply cannot answer and we need to let our teams know when a question is unanswerable." Deep dive: user-research-report_companion.md section 3 (Anatomy > Research Questions and Answers). ASK What was the question, in the plan's own words? What is the short answer, and which findings support it? If the plan stated a hypothesis, did the answer confirm or contradict it? Did the question stay open, and if so, why? PRIORITY One row per question the plan set, in the plan's own order. Add no new questions here; a pattern the study turned up that the plan never asked belongs in Key Findings instead. ROW HINT A good row restates the question, gives a short answer backed by a finding, and says plainly whether it confirmed a stated hypothesis or stayed open. A weak row answers a different, broader question than the one asked. GOOD | What makes someone miss the bulk-select control? | It reads as a filter toggle; 7 of 9 missed it on first use. | Confirmed the plan's hypothesis. | Answered | WEAK | How do people feel about duplicate photos? | They find them annoying. | | Answered | TRAP Summarizing the study's themes instead of answering the plan's own questions one by one, including a question nobody happened to raise in conversation but that the plan still asked. -->
| Research question (as the plan wrote it) | Answer | Confirms the plan's hypothesis? | Status ||---|---|---|---|| {{research_question}} | {{research_answer}} | {{confirms_hypothesis}} | {{question_status}} |
## What We Did
<!-- WHAT The method, the participants, and the dates, the limits of the sample stated as a fraction with its denominator, and one line on anything that ran differently from the plan. WHY Dovetail's own list carries "method and participants". NIST's template, inside the same EHR customization, structures the same content under "3 METHOD" and "3.1 PARTICIPANTS". UserTesting's detailed form carries "Methodology (e.g., unmoderated usability tests, surveys, etc.)". Sauro names the specific limit a small sample carries: "the key limitation is that you are limited to detecting large differences between designs or measures". The one-line deviation note is this library's own arrangement, the same device `experiment-readout` carries in its own Context and Setup section; no source instructs a report to note a deviation from its own plan. Deep dive: user-research-report_companion.md section 3 (Anatomy > What We Did). ASK What method did the study use, and does it match the plan? Who took part, stated as a fraction of those recruited? When did sessions run, and which version or state of the product were participants reacting to? What does a sample this size let you claim, and what does it not let you claim? Did anything run differently from the plan, and if so, what and why, in one line? GOOD "One-to-one moderated interviews with a walkthrough of each participant's own camera roll, 2026-03-10 to 2026-03-14. 9 of 10 recruited completed a session; one dropped for a scheduling conflict. A sample this size can surface a large, frequent problem; it cannot rank two close icon designs against each other. Ran as planned, except two sessions moved to phone calls when video access failed." WEAK "We talked to some users and took notes." (no method named, no participant count or denominator, no dates, no stated limit) TRAP Reporting a participant count with no denominator, so a reader cannot tell how many actually ran against how many were recruited. -->
- Method, and whether it matches the plan: {{method_used}}- Participants, as a fraction of those recruited: {{participant_summary}}- Dates sessions ran: {{session_dates}}- What this sample size lets you claim, and what it does not: {{sample_limits}}- One line on anything that ran differently from the plan: {{plan_deviation_note}}
## Key Findings
<!-- WHAT Each finding, stated as a pattern rather than a single anecdote, how many participants showed it as a fraction with its denominator, and the evidence, a quote or an observed action, a reader can check. Positive findings alongside problems, and a participant's own suggestion labelled as the participant's words. WHY NN/g names the chain this section sits at the end of: "Data leads to findings, which build insights, which justify recommendations." Dovetail: "A one-page summary with linked evidence beats a forty-slide deck nobody opens." Dovetail, more bluntly: "A finding without evidence is an opinion with formatting." Sauro and Lewis prefer fractions over a bare percentage: "Unless this information appears elsewhere in your report, we recommend presenting fractions." Schade: "Don’t forget to include positive findings as well, letting the team know what’s already working." GOV.UK: "Whenever you can, you should fully anonymise the extracts you use so that the participants cannot be identified." User Interviews names the suggestion row directly: "Any specific recommendations brought up by users during interviews or in surveys." Deep dive: user-research-report_companion.md section 3 (Anatomy > Key Findings). ASK What is the finding, as a pattern across sessions? How many participants showed it, as a fraction with the denominator stated? What evidence backs it, anonymised and within the consent given? Is this a problem, something that already works, or a participant's own suggestion? Have you included at least one thing that already works, not only what failed? PRIORITY Most important finding first. State counts as fractions with the denominator next to every count, never as a bare percentage unless the denominator already appears beside it. ROW HINT A good row names a pattern, gives a fraction with its denominator, and quotes or describes evidence a reader could go check. A weak row is an opinion with a number attached and no quote or observation behind it. GOOD | Bulk-select reads as a filter toggle, so people delete duplicates one at a time | 7 of 9 | "I thought that button was for sorting, not for picking more than one." | Problem | WEAK | Users don't like managing duplicates | Most of them | | Problem | TRAP Reporting an opinion with a number attached instead of a pattern with a quote or an observed action a reader could go check. -->
| Finding | Participants (fraction) | Evidence | Positive, problem, or participant suggestion ||---|---|---|---|| {{finding}} | {{finding_count}} | {{finding_evidence}} | {{finding_type}} |
## Task Results
<!-- WHAT For a usability test: each task's completion, errors and time against the measures the plan set, and the questionnaire scores the plan named. For any other method, one line saying so, with the table and the questionnaire line marked N/A. WHY This block's content comes from usability-specific sources only; no general report list carries it. NIST's template, inside its own health-records customization, structures it under "3.9 USABILITY METRICS" and "4 RESULTS", and defines a success rate directly: "The total number of successes were calculated for each task and then divided by the total number of times that task was attempted." UserTesting instructs a report to "Include quantitative data where applicable (e.g., success rates, charts, QX Score visualizations)". Questionnaire thresholds, where NIST's template states them inside the same EHR customization, are NIST's own convention, not a general usability bar this library sets. Deep dive: user-research-report_companion.md section 3 (Anatomy > Task Results). ASK Is this a usability test? If not, say so in one line and mark the table and the questionnaire line N/A. If so: did each participant complete each task, against the success criterion the plan set? How many errors, and how long did it take? What did any questionnaire score, and against what convention, if one applies? PRIORITY List tasks in the plan's own order. A success rate is the count that actually succeeded divided by the count that attempted the task, never a rate scored after the fact against a looser bar. ROW HINT A good row names the task, states completion as a fraction against the plan's success criterion, gives error and time figures where the plan collected them. A weak row gives a completion rate with no stated criterion behind it. GOOD | Find and merge the duplicate pair in this album | 8 of 9 | 1 minor | avg 42s | Plan's success criterion: merges without opening the help menu | WEAK | Duplicate task | 90% | | | | TRAP Reporting a completion rate with no criterion stated, so a reader cannot tell what counted as success. -->
- Method, if this study is not a usability test (the table and the questionnaire line are then N/A): {{not_usability_test_note}}
| Task | Completion (fraction, against the plan's criterion) | Errors | Time on task | Notes ||---|---|---|---|---|| {{task_name}} | {{task_completion}} | {{task_errors}} | {{task_time}} | {{task_notes}} |
- Questionnaire scores the plan named, and the convention they are read against, if any: {{questionnaire_scores}}
## Recommendations
<!-- WHAT What the team should do next, and why, with each recommendation tied to one or more findings above. It records no decision and names no chosen concept. WHY Dovetail's list carries "recommendations". User Interviews names "Recommendations" as its own section, and UserTesting prioritises it: "Prioritized based on impact". Schade's own instruction keeps the section honest about what it is: "But present recommendations as recommendations." The sources pull this section toward the decision itself: Dovetail: "End each finding with what the team should do about it, and who decides". Usercall: "If your report doesn't end in a decision or a debate about a decision, you haven't written a report, you've written a diary." Against that pull, this family's contract states the line this bundle follows: "A member here may recommend, but it records no decision." Deep dive: user-research-report_companion.md section 3 (Anatomy > Recommendations), section 6. ASK What should the team do next? Which finding or findings does this recommendation answer to? Why this change, rather than another? Have you stopped short of naming the decision itself, or the concept a team will build? PRIORITY Highest-impact recommendation first. Every recommendation names the finding it answers to; a recommendation with no finding behind it is an opinion, not a conclusion. ROW HINT A good row names a specific change, points at the finding it comes from, and says why that finding makes the change worth weighing. A weak row names a destination ("fix duplicates") with no finding and no reasoning behind it. GOOD | Make bulk-select the default view once duplicates are found | Bulk-select reads as a filter toggle (7 of 9) | Removes the step where people have to notice a control they currently miss | WEAK | Redesign the duplicates feature | | Seems like a good idea | TRAP Naming the concept the team should build, or stating that the team has decided to build it. That is a later document's job; this section names what is worth weighing and why. -->
| Recommendation | Tied to finding(s) | Why ||---|---|---|| {{recommendation}} | {{tied_finding}} | {{recommendation_rationale}} |
## Open Questions
<!-- WHAT What the study raised and did not answer, beyond the per-question status already recorded in Research Questions and Answers, and what research would answer it. WHY Dovetail's own list names "open questions" as its closing section. User Interviews names the device directly: "Compelling insights that illuminate unanswered questions". O'Neill describes closing a live readout the same way: "I'm like, okay, the next session needs to test these open questions. We still don't know X, Y, and Z." Userfeel asks for "further questions that arose from the test and how they could be investigated in the future". Anderson-Stanier names the underlying discipline: "we need to let our teams know when a question is unanswerable." Deep dive: user-research-report_companion.md section 3 (Anatomy > Open Questions). ASK What question did the study raise that it did not answer, whether or not the plan asked it? Why did it stay unanswered: wrong method, ran out of time, or outside this study's scope? What would a future study need to do to answer it? PRIORITY Order by how much a future decision depends on the answer, most consequential first. ROW HINT A good row names a specific, checkable question, says plainly why this study could not answer it, and names what a future study would need to do differently. A weak row is a vague area of curiosity with no path to an answer. GOOD | Do people delete duplicates in batches once they find bulk-select, or still one at a time? | The two sessions that found the control ran out of time before this came up. | A follow-up session that starts with bulk-select already visible. | WEAK | More research on duplicates in general. | | | TRAP Restating a question the study already answered, just more vaguely, instead of naming a question it genuinely could not reach. -->
| Question the study raised | Why it stayed unanswered | What would answer it ||---|---|---|| {{open_question}} | {{why_unanswered}} | {{what_would_answer_it}} |
## Appendix
<!-- WHAT The session list, the materials used, and where the raw notes live. Never a fixed retention period; that belongs to the organization's own standing policy. WHY Maze names the device plainly: "It's easy to overlook the appendix after putting together a detailed report, but all that glorious research data needs to be accounted for and referenced clearly." UserTesting's detailed form closes on "Appendices", carrying "Full responses, session details, additional data". NIST's template, inside its own EHR customization, closes on six named appendices, including "NON-DISCLOSURE AGREEMENT AND INFORMED CONSENT FORM". GOV.UK: "Keep your research data only in places that your organisation has approved for storing personal data." A fixed retention period belongs to a standing repository's own policy, not to this report's own figure. Deep dive: user-research-report_companion.md section 3 (Anatomy > Appendix). ASK Where is the session list, with dates and which participant attended which session? What materials were used (screener, consent form, task list, questionnaire)? Where do the raw notes or recordings live, and whose retention policy governs them? GOOD "Session list: linked, 9 sessions, 2026-03-10 to 2026-03-14. Materials: screener, consent form, and task list linked. Raw notes and recordings: in the shared research drive, under the org's standing 12-month retention policy, linked here, not restated as this study's own figure." WEAK "Notes are somewhere in the drive." (nothing a reader can actually open, no policy named) TRAP Restating the organization's retention period as though it were a figure this study set, rather than pointing to the standing policy that governs it. -->
- Session list, with dates: {{session_list_link}}- Materials used: {{materials_used_link}}- Where raw notes or recordings live, and whose retention policy governs them: {{raw_notes_location}}user-research-report_example.md
---title: "Discovery readout: Reporting friction study, 2026-05"study_name: "2026-05 Reporting Friction Study"author: "Priya Nair, PM, Reporting"plan_link: "../user-research-plan/user-research-plan_example.md"status: "final"last_updated: "2026-05-27"doc_type: user-research-reportsize: fullsource_template: user-research-reportsource_template_version: 0.1.0---
> **Worked example.** A filled `user-research-report`, full variant, for the same Acme Analytics product> used throughout this library's examples. Per the research-docs family contract's shared-scenario rule,> this document reports out the 2026-05 Reporting friction study that the> [research plan](../user-research-plan/user-research-plan_example.md) set out on 2026-04-20, and that four> other shipped examples in this library cite and, before this one, none of them shows: the> [Saved Views design brief](../design-brief/design-brief_example.md), the> [Saved Views PRD](../prd/prd_example.md), the [Saved Views epic](../epic/epic_example.md), and the> [Reporting squad sharing announcement](../announcement-internal-comms/announcement-internal-comms_example.md).> It is dated 2026-05-27, within the two weeks of the last session, 2026-05-15, that the plan promised, and> before the design brief of 2026-06-01 that first frames a concept from these findings, the PRD of> 2026-06-12, and the epic of 2026-06-13. Nothing in the body below names a concept, a feature, or a product> decision; this report answers the plan's own questions and recommends, and leaves framing and deciding to> the documents that follow it.>> All figures are illustrative. They are internally consistent with the other examples in this library and> are not drawn from a real company.
# Discovery readout: Reporting friction study, 2026-05
## Summary
- What the study found, and what it recommends: Re-filtering, not exporting or sharing, is what analysts repeat most on every visit to their dashboard. 9 of 12 described rebuilding filters by hand several times a day, and 3 already keep a personal text file of "the filters I always use" to avoid relying on memory. We recommend funding work so that an analyst no longer rebuilds the same filters on every visit, and weighing whether that work should reach a teammate too.- Decision these findings inform, and who makes it: Whether the Reporting team funds work next quarter on letting analysts avoid rebuilding their filters from scratch, a decision the Reporting team's own leadership makes.
## Background and Objectives
- Why this study ran: A request to let the dashboard hold onto an analyst's filters between visits had come up twice in the past year and been deferred both times, each time for lack of more than a handful of comments that it might be worth the engineering time. The January user persona interviews raised a version of the same friction independently, across a different set of accounts than either request came from, which was enough overlap to check properly before the request came up a third time.- Decision it informs, restated for a reader who has not seen the plan: Whether the Reporting team funds work next quarter on letting analysts avoid rebuilding their filters from scratch.- Link to the plan: [2026-05 Reporting Friction Study Research Plan](../user-research-plan/user-research-plan_example.md)- If this studies something people already use, why it is still discovery work: This studies Acme's dashboard as customers use it today, not a concept that does not exist yet. It is still discovery work because the study looks for where an opportunity exists inside something already shipped, not for whether the dashboard meets a specification it was built against.
## Research Questions and Answers
| Research question (as the plan wrote it) | Answer | Confirms the plan's hypothesis? | Status ||---|---|---|---|| Where in a session does an analyst's time go to rebuilding something rather than reading what the dashboard already shows? | To rebuilding filters. Re-filtering was the single most frequent repeated action in the dashboard, ahead of export and share. | Confirmed | Answered || When an analyst rebuilds filters after the dashboard resets to its default, how often does what they rebuild actually match what they meant to be looking at? | Not always. Analysts described filters set incorrectly from memory, which risks an analysis of the wrong slice (Key Findings, row 2). The sessions did not count how often it happened. | N/A | Answered || What do analysts already do, inside the product or outside it, to avoid rebuilding filters from memory? | 3 of 12 already kept a personal text file listing "the filters I always use"; the rest rebuilt from memory each time. | N/A | Answered || Does this differ between an analyst who opens the dashboard several times a week and one who opens it about once a week? | Could not be answered from this study. See Open Questions, below. | N/A | Open || How often does rebuilding filters come up in a session compared with other repeated actions, like exporting a view or sharing one with a teammate? | More often than either. Re-filtering outranked both exporting and sharing as the most frequently repeated action logged across the sessions. | N/A | Answered |
## What We Did
- Method, and whether it matches the plan: One-to-one moderated interviews built around a walkthrough of the analyst's own live dashboard, about an hour each, matching the plan. Participants reacted to the dashboard as it shipped at the time. No prototype or mockup was shown, since nothing had been designed yet.- Participants, as a fraction of those recruited: 12 of 15 recruited completed a session; the other three, held in reserve, were not needed.- Dates sessions ran: 2026-05-04 to 2026-05-15, with a pilot session on 2026-05-01.- What this sample size lets you claim, and what it does not: A study of this size is good at surfacing a problem shared by most of a screened group; it is not built to rank two slightly different ways of saving a filter set against each other. The sessions covered only the small set of report types these analysts happened to use, so nothing here shows the finding holds for every report type Acme ships.- One line on anything that ran differently from the plan: Nothing. Sessions ran exactly as the plan described, including the pilot.
## Key Findings
| Finding | Participants (fraction) | Evidence | Positive, problem, or participant suggestion ||---|---|---|---|| Re-filtering was the single most frequent repeated action in the dashboard, ahead of export and share. | All 12 sessions, counted together | The note taker's log of repeated actions, totalled across the twelve sessions, as the plan set out. | Problem || 9 of 12 analysts described manually rebuilding filters several times a day. | 9 of 12 | "Every time it resets I'm building the same filters again, and every so often I get one wrong and don't notice until the numbers look off." | Problem || 3 of 12 already kept a personal text file listing "the filters I always use," so they would not have to rebuild from memory. These files were the only saved setups in the study, and no teammate could reach them. | 3 of 12 | Each of the three showed the moderator the file on their own screen partway through the walkthrough. | Problem || Analysts found and used Acme's existing filter controls without help; the trouble was remembering which filters to set, not setting them. | 12 of 12 | No participant asked the moderator how to apply a filter during the walkthrough. | Positive |
## Task Results
This study was not a usability test. Each session was a walkthrough of the analyst's own dashboard andtheir own data, not a set of tasks handed to them, so the table below and the questionnaire line are bothN/A.
| Task | Completion (fraction, against the plan's criterion) | Errors | Time on task | Notes ||---|---|---|---|---|| N/A | N/A | N/A | N/A | Not a usability test; see the note above |
- Questionnaire scores the plan named, and the convention they are read against, if any: N/A, not a usability test.
## Recommendations
| Recommendation | Tied to finding(s) | Why ||---|---|---|| Fund work next quarter so that an analyst no longer rebuilds the same filters on every visit. Leave how to the design work that follows. | Re-filtering is the most frequent repeated action in the dashboard; 9 of 12 rebuild filters by hand several times a day; 3 of 12 already keep a personal text file to avoid it. | Removes the repeated action the study logged most often, and the risk that a filter set from memory is wrong and shows the wrong slice of data. || Weigh, in that same work, whether a teammate should be able to start from the filters someone else set up. | The only saved setups in the study were personal text files no teammate could reach; some participants who lead a team raised the same want (Open Questions). | Each analyst today rebuilds their own version of a shared slice, so two people can read different numbers for what they believe is the same view. |
## Open Questions
| Question the study raised | Why it stayed unanswered | What would answer it ||---|---|---|| Does re-filtering cost an analyst who opens the dashboard several times a week more time than it costs one who opens it about once a week? | Recruiting screened every participant only for the Recurring Analyst's minimum weekly cadence, not for this finer split, so the twelve sessions cannot be reliably divided along that line. | A follow-up study that screens and recruits specifically on opening frequency, rather than on the broader Recurring Analyst definition. || How many team leads want their team looking at one agreed view, rather than each person's own reconstruction of it? | Some participants who lead a team raised this unprompted, but none of the plan's questions asked about it, so the sessions cannot say how common the want is. | A future study that asks team leads directly whether they would hand their team one agreed view. |
## Appendix
- Session list, with dates: Linked in the UX Research function's shared drive: 12 sessions, 2026-05-04 to 2026-05-15, one row per participant with the date their own session ran.- Materials used: The screener used to recruit the twelve participants, linked in the same shared drive. No separate consent form; the spoken statement described in the plan's Consent, Privacy and Ethics section is what every session used.- Where raw notes or recordings live, and whose retention policy governs them: In Acme's standard research storage, visible only to the UX Research function and the Reporting teammates who observed a session, under Acme's standing research-data retention policy, linked in the UX Research function's internal space, without restating that policy's length as a figure this report sets.Provenance
Section titled “Provenance”The reasoning, the history and every source, in the repository:
- Companion - the long-form argument: why these sections, where the sources disagree, and what the bundle refuses to claim
- History - what changed in this bundle, and when
- Research log - every source consulted, with what each one actually supports
- Catalog metadata - the machine-readable record this page is generated from
Catalog record: 9 sections across 1 format(s), methodology design thinking/dual-track, typically owned by UX Researcher.