
Redesigning reporting to reduce churn
Performance at a glance, details on demand.
One dashboard shows overall purchases alongside affiliate-specific transactions, each split by source with the change from the last period. Purchase attribution lives in a table managers can search, sort or export themselves, no ticket required.
01
Purchase summary
Two cards surface total revenue and affiliate-driven revenue side by side, changing dynamically to the selected date range.

02
Trends over time
One chart overlays total purchases and sales with their affiliate-driven counterparts, so the program’s contribution is always visible in context.

03
Self-serve export
Exports used to mean a help desk ticket. Now it lives in the Purchases table and opens pre-filled with the screen’s date range.


Better data changed things for everyone.
Program managers finally had what they needed to run their programs with confidence. Support reclaimed time that was going toward manual data pulls. And for the business, a feature that was driving cancellations became something customers pointed to as value.
No more guessing
Managers had a clear view of how their program was performing and could pull the data they needed on their own.
Close the churn gap
Closed a top churn driver and gave support and engineering time back that was going toward manual data pulls.
A cleared queue
Exports stopped routing through the help desk, freeing the team for problems that needed a person.
“This is so much more than I envisioned.”
- Affiliate program manager, beta user
Churn was creeping up month over month.
The Purchases screen gave managers a line graph and not much else. They couldn't export their data or see how individual affiliates were performing, so they kept asking support to pull it for them. The backlog grew, and managers who couldn't prove the value of the platform were the ones most likely to leave.

Where frustration was loudest: the help desk.
The support tickets told me who to talk to, the interviews helped me understand why they were frustrated and what they needed, and an in-app survey told me how widespread it was. Each step set up the next.
Help desk tickets
The support team had already noticed a pattern with reporting complaints. I filtered tickets for data and reporting issues, looked for recurring themes, and used that list to recruit interview participants.
User interviews
I designed and led five moderated interviews end to end. I wrote the script, built the screener, did outreach, scheduled sessions, and moderated with my PM taking notes. Participants were segmented by account age, support ticket history, and annual revenue generated through the platform.
In-app survey
While interviews were running, I put a single intercept survey question on the old reports page to pressure-test what I was hearing. We ran it until we hit around 100 responses. The results confirmed that reporting frustration wasn’t isolated to the users we interviewed.
“Reporting, reporting, reporting. It will change the world.”
— Matt, Affiliate program manager
Designing around the right questions
The reporting suite spanned 8 screens. Working with the PM, I mapped each screen to a core question, pairing KPIs, metrics, and actions a user would need to be able to answer questions about their program performance.
How much revenue are my affiliates generating?

Anchoring design decisions to a single question kept each screen focused and prevented scope creep. Throughout the process, I kept engineering and design counterparts in the loop, running team reviews to pressure-test the feasibility of each data point. Those reviews are where the screens took their final form.
Aligning on what the MVP needed to be.
Our early concepts were ambitious. Six KPIs bucketed into a configurable chart, a Top Affiliates leaderboard, and full filtering capabilities. But after working closely with engineering and product to gut-check viability and effort, we made a deliberate call: ship the version that answered the core question and roadmap the rest. Three constraints shaped what made the cut:
The full feature set was too much to build and ship in a single release.
Less complexity meant less that could go wrong in beta.
Shipping the core first let us validate before investing in the rest.

Hover or tap to Highlight
Simplified from six options to two cards split by source
Fixed to purchases and sales instead of a configurable axis
Moved off the Purchases dashboard and into the Affiliates screen
Deferred to the roadmap until core usage was validated
In hindsight
Map data before mapping screens
I’d spend more time upfront working through what data we actually had, what was missing, and what was buildable. Information architecture for reporting isn’t just about layout. It’s about knowing what you can promise the user before you design it.
A linear process is rare
Research led directly into strategy, which led into product roadmap. That rarely happens. It did here because the team trusted the process and gave me room to shape it. That’s something I’d try to protect on every project going forward.
Keep the feedback loop
Export generation could take time depending on report size, and async delivery wasn’t feasible. So we designed a progress state that kept users oriented while they waited. Technical constraints don’t always mean a worse experience. Sometimes they just mean a different design problem.
