Our Research Repository Had 400 Reports and Almost No Readers
We stopped treating research as an archive and started treating it like a product, with users, formats and a maintenance plan.
Two years ago, our team had what looked like an impressive research repository. Over 400 reports, neatly tagged, going back five years. When I checked the analytics, the average report had been opened 1.3 times. Usually by the person who wrote it.
It was not a knowledge base. It was a very well-organised graveyard.
Why repositories die
When I interviewed product managers and designers about why they did not use it, the answers were consistent:
- I don't know what's in there, so I don't think to look.
- Reports are too long. I need the answer, not the method.
- Half of it is outdated and I can't tell which half.
- It's faster to just message a researcher.
None of these were about the tool. They were about how we packaged and maintained what went into it.
What we changed
1. Insights, not reports, became the unit
We started breaking each study into individual insight cards. Each card had one finding in a single sentence, the evidence behind it, the date, and a confidence level. A single study might produce eight cards. People search for answers, not documents, so we stored answers.
2. Every card got an expiry date
Each insight now has a review date, usually twelve months after it was created. When a card expires, the owning researcher either confirms it, updates it, or archives it. This one rule did more for trust than anything else. People finally knew that what they found was current.
3. We went to where people already were
Instead of expecting people to visit the repository, we pushed insights out. A short fortnightly digest in the team channel with three relevant cards. A habit of referencing a card in any planning document that relied on it. Within three months, most repository visits came from those digests.
4. We tracked it like a product
We set simple metrics:
- Weekly active readers
- Cards cited in planning documents
- Percentage of cards past their review date
The third number started at 71 percent. Today it sits under 15.
The results, honestly
Usage is not huge, but it is real. Around 40 people read at least one card each week, out of a product team of about 70. More importantly, planning documents now reference prior research before new studies are proposed, which has prevented at least four duplicate studies this year.
A repository is only as useful as the trust people have in its freshness.
If your research archive feels ignored, resist the urge to switch tools. Start smaller. Break findings into single claims, give them an expiry date, and bring them to the places your colleagues already spend their day.
Cover: Photo by Luke Caunt on Unsplash.
