Making research findable: service blueprint
Context: research operations at a large national retailer
I owned the research repository and rebuilt how teams found and requested research, so they could build on existing knowledge and make informed decisions. Below: the service before and after the redesign.
- The problem
- Research was hard to find, so teams waited on one person's inbox, repeated studies that already existed, lost track of past studies, and made decisions on partial evidence.
- The change
- Teams could look in one shared place first, ask a chat assistant, and send one request form that routed itself.
- Why it matters
- Less time searching and waiting, no paying twice for the same research, better-informed decisions, and customers asked only what is still unknown.
Three principles that guided my work.
- Lighten the load
- A service that depends on one person's inbox doesn't last.
- Automated intake and routing so the team stopped acting as order-takers.
- As an interim fix, set up shared access so no one lost it for being inactive.
- Delivered at no added cost to the organization.
- Connect and listen
- Ran surveys to hear what people thought of the tool.
- Acted on the answers by moving to one shared place.
- Connected people working on similar problems across teams.
- Built the request form together with the design ops team.
- Go where people are
- Asking people to come to a tool didn't work.
- Took research to where people already worked.
- Used a chatbot in team chat and AI-assisted request forms to do it.
One request, followed through five steps from three points of view.
Every lane for the request flow, from the requester's first question to the work behind the scenes.
An illustrative reply from the research assistant, showing the fields every study carried.
Interviews on how people narrow down choices and what information they trust.
Tested three comparison layouts with first-time and repeat shoppers.
Why this format
- Who did it, and their team, so you can contact the researcher and see who's working on similar problems.
- A clickable title that opens the full readout.
- The date, so you can judge how current it is.
- A short summary, so you can decide quickly if it's relevant.
- A “Follow this topic” button, so you stay up to date on what matters to you most without having to ask again.
It mirrors the repository's standard template, so every study carried the same fields wherever it was found.
What the old method cost, and what the redesign was built to change. This is the reasoning behind the work, and measuring the results is the next step.
Time
Finding research meant searching separate parts of one tool, asking around, or waiting for a person to reply. Looking first, in the chat people already use, shortens that wait.
Money
When teams can't see what exists, studies get repeated, and the company pays twice to learn the same thing. Finding existing research first keeps new research for questions that are still open. The work itself added no cost for the organization.
Better decisions
A decision made without the full picture rests on partial evidence. With past findings easy to reach, teams start from what is already known and move faster with more confidence.
Visibility and collaboration
Before, it was hard to know what was happening, where, and who else was working on similar problems. Seeing who did each study makes it easier to connect and share work.
The research team's time
A small team spent its hours uploading studies, sorting requests and making introductions by hand. Automation and self-service free that time for reviewing plans and doing research with teams.
Care for users
Better-informed teams build on what customers have already told us instead of asking again, so each new conversation with a customer goes toward something not yet known. Much of the research and design behind this work was about store associates, to make their processes easier.
About this page: Illustrative recreation. The personas, thoughts and feelings are written to show the experience, not quotes from real people. The chatbot's studies, names, teams and dates are invented; its reply format is real. Tool names are generalized.