Coffee Chat Operations Dashboard

LONG STORY SHORT

CONTEXT

The program was growing faster than the process

MSTI's Coffee Chat Program connects prospective students with current MSTI ambassadors for one-on-one conversations about the program. At its peak, coordinators were managing 100+ requests per semester across a SharePoint spreadsheet, manual email chains, and a Microsoft Form for post-chat feedback.

The program worked but the workflow behind it was held together manually. Every step from assignment to feedback collection required someone to copy, paste, update, and follow up by hand. During the program's busiest weeks, things slipped through.

My role was not to rebuild the program from scratch. It was to observe the existing workflow, find where the team was losing the most time, and determine whether a lightweight tool could recover it.

DISCOVERY

Observe first, build second

Before proposing any solution, I ran 8 structured stakeholder interviews over two weeks 6 with student ambassadors and 2 with faculty coordinators. Each session followed the same prompt: "Walk me through what you actually do when a new request comes in, step by step."

I was specifically listening for three things: where time was being lost, where ownership was unclear, and what tasks felt repetitive enough that people had developed workarounds. I wasn't leading toward a solution I was mapping reality.

Student intake form on the MSTI website. Each submission became a row in the shared tracker.

Original SharePoint spreadsheet with no ownership column, no status pipeline, and no SLA visibility.

Three friction points surfaced consistently across nearly every interview:

WORKFLOW MAPPING

Turning observations into a process map

After the interviews, I mapped the full end-to-end workflow in FigJam both the coordinator flow and the ambassador journey to make the friction visible to everyone, not just to me. The goal was to give the team a shared picture of reality before we discussed what to fix.

The existing process had 9 steps from form submission to feedback collection. Three of them highlighted in the map were purely manual and offered no visibility to anyone else:

FigJam workflow map red nodes indicate manual, high-friction steps with no visibility to coordinators

The FigJam map also surfaced something the interviews alone didn't fully capture: the ambassador journey was almost entirely invisible to coordinators. Once an assignment was made, there was no signal about whether the ambassador had reached out, whether the chat had happened, or whether anything had gone wrong until the feedback form was manually sent a week later.

This became the core design constraint: the solution needed to give coordinators real-time visibility into ambassador actions without changing how ambassadors worked.

SCOPING

Deciding what to build and what to leave out

With the friction map in hand, I evaluated each pain point against three questions: How much time does this cost? Can it be solved without a backend? Will the team actually use it?

I scoped the prototype around three targeted interventions one per major friction point and explicitly left out anything that would require replacing the existing SharePoint workflow or adding backend infrastructure. The constraint wasn't technical. It was adoption: the team needed to see value within days, not weeks.

BUILD & SHIP

Building and deploying the dashboard without an engineering team

Rather than writing a requirements doc and waiting for an engineering sprint, I built the prototype myself using AI-assisted tooling. The approach: describe the interface, iterate on the output, validate with users, repeat. This let me put something real in front of stakeholders within days of scoping not weeks.

The prototype is a two-view web dashboard. The coordinator view gives program staff a real-time pipeline of all requests status, days in stage, stall flags, and ambassador assignment replacing the need to manually read the spreadsheet. The ambassador portal gives student ambassadors a personal queue with pre-filled email templates and a built-in post-chat feedback form, replacing the manual Outlook copy-paste workflow.

Coordinator view request pipeline with status badges, stall flags, and ambassador assignment

VALIDATION

Deployed and adopted by the team

Once the prototype reached a functional state, I ran it through two rounds of testing. The first was a walkthrough session with 3 ambassadors I asked them to complete a set of tasks (find a request, send an intro email, submit feedback) without instruction, and observed where they hesitated or got confused. That session surfaced two friction points in the ambassador portal that I fixed before the wider rollout.

The second round was a structured survey distributed to all 8 stakeholders from the interview cohort after a demo session. Based on the feedback, I iterated on the interface and deployed the dashboard to Vercel. It is now actively used by 8 MSTI faculty members to manage coffee chat requests across the semester.

LEARNINGS

What I'd do differently