UX and research · Intermediate · 15 min
Keep research findings in a searchable repository
By Padding · Updated 18 Nov 2026
Any spreadsheet (Google Sheets, Excel or Numbers) · November 2026
You will have a small spreadsheet repository holding every finding from one study, each row tagged by topic, study and date and linked to its evidence, plus a tag list and house rules a colleague can follow without asking you.
On this page
Six months after a study, someone asks "did we ever learn why people abandon the repeat prescription form?" The answer exists, somewhere in a slide deck, a folder of transcripts and the memory of a researcher who has since moved team. So the question gets researched again, or worse, guessed.
A research repository fixes that by keeping findings, not files, in one place you can search. Most attempts fail for the opposite reason to the one people expect: not because the tool is too weak, but because the setup is too heavy. A dedicated platform, forty tags and a mandatory template for every note will be tidy for one quarter and abandoned by the next.
This tutorial builds the smallest version that still works: one shared spreadsheet with four tabs. One row is one finding. Every row says which study it came from, when, what it is about, and where the evidence lives. You will load it with the findings of one study and write down the rules so the next person adds rows the same way.
The worked example is a fictional study: five usability sessions on the repeat prescription flow of a pharmacy app, run in October 2026. Replace it with your own study as you go.
Step 1 · Decide what one row holds
Before you open the spreadsheet, settle the unit. A row is one finding: a single claim that stands on its own when read out of context, months later, by someone who was not in the study.
That rules out three things people often try to store instead:
- Raw observations: "P2 tapped Back twice on the dosage screen." Useful as evidence, too small to search for. They stay in your notes and the row links to them.
- Whole reports: a link to a 30-page deck is a filing system, not a repository. Nobody can filter it.
- Recommendations on their own: "Add a progress bar." Without the finding behind it, a later reader cannot tell whether it still applies.
A good row reads like this: "People who reorder more than one medicine lose track of which items they have already added, because the basket is hidden until the last screen."
Take your study's findings and rewrite any that fail the test. If a finding only makes sense next to the slide before it, it needs more words. If it covers two separate behaviours, split it.
Check it worked
You have a plain list of findings, each one sentence or two at most. Read three of them aloud without any surrounding context. Each should make sense to a colleague from another team, and none should contain "this", "the above" or "as shown".
Step 2 · Build the sheet with four tabs
Create a new spreadsheet and name it so it sorts first in your shared drive, for example "Research repository". Add four tabs:
- Findings: one row per finding
- Studies: one row per study
- Tags: the allowed tag list
- Read me: the rules, which you write in the last step
On the Findings tab, add these nine column headers in row 1, in this order:
- ID: a short unique code, such as F-001
- Finding: the one-sentence finding
- Study: the study code from the Studies tab, such as S-2026-07
- Date: the month the study ran, as YYYY-MM, such as 2026-10
- Topic tags: one to three tags from the Tags tab, such as basket, reordering
- Strength: how many participants showed it, written as "4 of 5"
- Evidence: links to the notes, clips or quotes behind it
- Status: Current, Superseded or Retired
- Added by: the initials of whoever added the row
Keep it to these nine columns. Every extra column is one more thing someone skips when they are busy, and an empty column makes the whole sheet look unreliable.
Freeze the header row so it stays visible as the sheet grows.
Check it worked
Your spreadsheet has four tabs with the names above. The Findings tab has nine headers in row 1, and when you scroll down the headers stay in view.
Step 3 · Write a short, controlled tag list
Tags are what make the repository searchable, and they are also what usually ruins it. If one person tags "checkout", another "payment" and a third "basket-flow", a filter finds a third of the evidence.
On the Tags tab, make two columns: Tag and Use it for. Write between eight and fifteen tags. Base them on parts of the product and on recurring user goals, not on the themes of one study, because themes change every study and product areas do not.
For the pharmacy example:
- sign-in: logging in, passwords, account recovery
- reordering: finding and repeating a past prescription
- basket: reviewing and changing items before submitting
- collection: choosing a pharmacy, delivery or pick-up
- notifications: reminders, ready-for-collection messages
- trust: worries about privacy, accuracy or safety
- accessibility: barriers for disabled people or older devices
- language: wording, labels, medical terms
Write tags in lowercase, use a hyphen instead of a space, and give each one a single meaning. If two tags keep turning up together, merge them.
Now limit the Topic tags column to that list. In Google Sheets and Excel you can turn a column into an in-cell dropdown that reads its options from a range; point it at the Tag column on the Tags tab. Google Sheets dropdowns can allow more than one choice per cell. If your spreadsheet only allows one, use three narrow columns (Tag 1, Tag 2, Tag 3) instead of one.
Check it worked
The Tags tab lists eight to fifteen tags, each with a one-line description. Click an empty cell in the Topic tags column on the Findings tab: it offers your tags and nothing else, and typing a tag that is not on the list is rejected or flagged.
Step 4 · Log the study
On the Studies tab, add these headers: Study code, Name, Dates, Method, Participants, Researcher, Plan, Report.
Then add a row for your study. For the pharmacy example it holds:
- Study code: S-2026-07
- Name: Repeat prescription usability
- Dates: 2026-10-06 to 2026-10-10
- Method: moderated remote usability test
- Participants: 5 adults who reorder monthly, recruited from app users
- Researcher: AO
- Plan and Report: links to the documents
Use a code made of the year and a running number. It is short enough to type into every finding row and still tells a reader roughly how old the study is.
The Participants column describes who took part, never who they were. Names, email addresses and phone numbers do not belong anywhere in the repository. The people you interviewed agreed to take part in a study, not to be listed in a document that the whole company can open.
Check it worked
The Studies tab has one row with all eight cells filled. The Plan and Report links open for a colleague who has access to the shared drive, and no cell contains a participant's name or contact details.
Step 5 · Enter the findings with links to evidence
Copy each finding from your list into the Findings tab, one per row, and fill every column. The column that takes the most care is Evidence.
A link to the whole transcript folder is not evidence; it is homework. Link to the smallest thing that proves the point:
- A timestamped clip, or a recording link with the time noted, such as "P3 at 14:20"
- A specific note or quote in your synthesis board or document
- The row in your evidence sheet, if you kept one while writing insights
Put two or three links per finding, from different participants where you have them. If you can only find one, the Strength column should say so ("1 of 5"), and the row is still worth keeping: a weak finding that is labelled as weak is more useful than one that is lost.
Write the Strength as a count, not a label like "high". "4 of 5" means the same thing to everyone. "High" means something different to each reader.
For the pharmacy example, the first three rows hold these values (ID, study, date, tags, strength and status, then the finding):
- F-001 · S-2026-07 · 2026-10 · basket, reordering · 4 of 5 · Current: People who reorder more than one medicine lose track of what they have added, because the basket is hidden until the last screen.
- F-002 · S-2026-07 · 2026-10 · trust, notifications · 3 of 5 · Current: People do not trust that a reorder has been sent until they get a message, so several resubmitted it.
- F-003 · S-2026-07 · 2026-10 · language, reordering · 3 of 5 · Current: The label "Repeat" is read as "repeat the last order exactly", so people avoid it when one item has changed.
Check it worked
Every finding from your study has a row, and no cell in the Findings tab is empty. Pick any row and open its first evidence link: within a minute you land on the exact moment or quote that supports it, not on a folder or the start of a long recording.
Step 6 · Test it with someone else's question
A repository is only searchable if it answers questions you did not plan for. Ask a colleague who was not in the study (a developer or product manager works well) for a question they have about the product. If nobody is free, use one of these:
- "What do we know about trust?"
- "Have we seen problems with the basket?"
- "What did we learn in October?"
Turn on a filter for the Findings tab and answer the question using only the filter: by Topic tags, by Date, or by Study. Then read the matching rows to your colleague.
Watch for three failures. A question that finds nothing, although you know the answer is there, means a tag is missing or the finding is filed under a different one. A question that finds rows your colleague cannot understand means the finding text leans on context they do not have. A question that returns the right row but no usable evidence means the links are too broad. Fix each one in the sheet now, while the study is fresh.
Check it worked
You have run at least two questions through the filter. Each returned the rows you expected, your colleague understood them without you explaining the study, and you made any fixes the test showed up.
Step 7 · Write the house rules and the upkeep routine
The rules are what let the repository survive you. On the Read me tab, write short answers to these questions:
- What is a finding? One self-contained sentence, backed by at least one evidence link.
- Who adds rows? The researcher or designer who ran the study, within two weeks of it ending.
- How do I tag? One to three tags from the Tags tab. To propose a new tag, add it to the Tags tab with a description and tell the owner.
- What if a newer study contradicts an older finding? Set the old row's Status to Superseded and add the new finding's ID in its Finding cell, for example "Superseded by F-031". Never delete rows; the history of what you believed is part of the evidence.
- When is a finding Retired? When the part of the product it describes no longer exists.
- Who owns it? One named person, who reviews the Tags tab once a quarter and merges or removes tags that are unused or duplicated.
Keep the whole tab short enough to read in two minutes. Then add one line to your own study checklist, after "share the report": "Add findings to the repository." That line, more than any tool, is what keeps the sheet alive.
Share the spreadsheet with your team so everyone can view it and the people who run studies can edit it.
Check it worked
The Read me tab answers all six questions and names an owner. Ask the colleague from the previous step to add a made-up finding using only the Read me tab. They fill all nine columns, use existing tags, and do not need to ask you anything. Delete their test row afterwards.
Next steps
Your repository now holds one study's findings in a form anyone can filter. Add each new study the same way, and resist new columns until the team has asked for the same thing three times.
- Write sharper findings before they go in: Turn research themes into insight statements
- Get the team to act on what the repository holds: Report research findings so the team acts
- Build the evidence that rows link to: Build an affinity map from interview notes
- Make sure participants agreed to how their data is kept: Get informed consent for user research
Part of these paths
Related tutorials
- Report research findings so the team acts on themYou will have a short findings report that opens with the decisions it supports and backs each finding with evidence, plus a 15-minute playback plan that ends with a named owner for each recommendation.
- Turn research themes into insight statementsYou will have five insight statements, each saying what people do, why they do it and what it means for the design, with quotes from several participants and a confidence level beside it.
Need this done for you?
Padding does this work for growing companies, from the first diagnosis to a system that keeps working.
Get new tutorials