Case study ·
CS-03
·
Minerva
Retiring 400+ hours of manual reporting a year with one dashboard
Forum’s first in-app data dashboard, for instructors and partners. I found the problem, then led it as both designer and PM.
01
Rich classroom data, and almost nobody could reach it
Forum is a science-based online learning platform used by universities around the world to create, deliver, and assess pedagogically sound curriculum. Every class session generates engagement data: student and instructor talk time, chat and reaction activity, attendance, breakout participation. Then it all disappeared after class, locked in backend databases and third-party analytics tools, unintelligible without an engineer.
So Partner Success bridged the gap by hand: data-heavy engagement reports, per partner, every term. Slow to build, delivered too late to act on, and often unread. As the partner base grew, the math only got worse. Nobody handed me this project; I saw both a scalability issue and a strategic opportunity, and pitched replacing the reports with a live, self-serve dashboard, in the app, where class happens.
Field note
A report nobody reads, rebuilt by hand every semester, is a product asking to exist.
02
The questions pointed at a product, not better reports
I started by meeting the customer-facing stakeholders and instructors: how were the reports built, and what did partners actually value in them? The production process wasn’t scalable, but the reports encoded the key user goals. Everyone was trying to answer three questions: How is this program or course doing overall? Are any sections struggling? Which students might be at risk, and how soon can we know?
Those questions don’t want a prettier PDF. They want insights over time, at the program, course, section, and class level. That reframe turned a reporting cleanup into a product.
The tell
“How soon can we know?” is a monitoring question. Nobody asks a static report how soon.
How might we make engagement data more useful and accessible to partners, while still aligning with our pedagogical standards?
03
Users wanted one simple score. We couldn’t honestly give one.
Three calls shaped the dashboard, and the middle one, the refusal, is the one this case study exists for.
Honest ledger
The score would have demoed beautifully. It also would have been wrong often, with authority.
One number would have been simpler. It also would have been wrong.
What shipped instead — signals, ranked by trust: 01 talk time (primary, format-dependent) · 02 reactions & hands (secondary) · 03 chat (context) · 04 attendance (context) · 05 window focus (weakest signal). Shown in that order. The instructor interprets; the design never issues a verdict.
04
Built to answer “who do I focus on next class?” in two seconds
The final dashboard gives instructors three key metrics up front: student talk time, instructor talk time, and attendance, each with a colored health bar so engagement reads at a glance. Talk time sorts lowest-first, so the students who need attention rise to the top, and “good” is a deliberately wide band, so the dashboard doesn’t raise false alarms. Below sit the supporting signals (polls, chat, breakouts, reactions) and talk-time history for the last four classes, so one quiet session reads as a data point, not a verdict.
In miniature
Color communicates health; the number underneath carries the nuance. That split is DL-02, shipped.
05
400+ Partner Success hours back, every year
Those hours are Partner Success report-production time: the dashboard removed the need for most manual reports, with the remainder scheduled on the roadmap. It shipped to all 20+ partners as Forum’s first in-app dashboard (and earned a Minerva blog feature), and the team’s freed hours moved to demos, support, and upsell conversations.
“The engagement chart is everything I’ve ever wanted.” — Instructor, Forum partner
“These are super helpful for personal reflection and faculty reviews!” — Instructor, Forum partner
06
What changed beyond the interface
The quietest outcome mattered most to me: full alignment with Minerva’s internal academic team, whose approval was critical given their high pedagogical standards. Instructors began using the data to inform their in-class decisions, which is exactly the reflective use the SME and I designed for.
Product direction, self-initiated
Nobody handed me this project. I saw both a scalability issue and a strategic opportunity, and pitched replacing the hand-built reports with a live, self-serve dashboard — then led it as both designer and de facto PM.
07
The best feature is the score we refused to ship
Instructors wanted a single score, and the concept test proved how much: the direction carrying one won 91% of the room. We refused to ship it anyway, because we couldn’t compute it honestly, and traded that simplicity for signals ranked by trust. I’d make the trade again. The pattern travels anywhere data meets people: decide what you can responsibly say before deciding how to say it.
There’s an epilogue, too: the dashboard now generates exactly the data an honest score would need. The refusal was a deferral, not a forever.
Closing note
Two months, two collaborators, four hundred hours a year. Small teams can retire big drudgery.