The $5.8M save nobody wanted to make a slide about.

The problem

Design at Panorama had a seat at the table but didn’t have a vocabulary the C-suite respected. Quarterly business reviews ran on instinct. Leadership asked the same question every planning cycle: where is design’s measurable contribution? The honest answer was a list of NPS deltas and a link to the design system, which didn’t move a balance sheet.

The thing the org needed was a way to take a single user interaction — a click, a session, a feedback moment — and translate it to dollars the leadership team would recognize. I called it the Customer Value Index (CVI). CVI was the metric. UX Lite was the instrument.

The process

I started with the data we already had. Support tickets, churn-interview transcripts, product analytics, CSM QBR notes. None of it was structured for design. All of it was sitting in a Slack thread or a Sheet somewhere.

The first version of CVI scored each account on a small set of UX indicators I could pull without engineering involvement: error-rate bucket, feature-depth bucket, accessibility issues bucket. Coarse. Directionally useful. Cost nothing to ship because the data was already there.

UX Lite was the survey instrument: a 90-second, three-question weekly sample sent to a slice of users. Built it as a Typeform, then graduated it to an in-product surface because the sample needed to feel like a real product, not a survey. The technical lift was small. The vocabulary shift was enormous — for the first time, design owned a weekly signal that the CEO could read.

The second version of CVI (V2) layered that weekly signal onto the account-level churn model. After six weeks of cross-correlation, we could see, with real numbers, which account cohort was decaying fastest and why. The output was a dashboard and a quarterly narrative — but the narrative was what changed behavior.

The outcome

Two quarters in, the dashboard caught a $5.8M ARR account at risk eleven weeks before the renewal conversation caught it. The issue: a major onboarding flow had drifted out of compliance with the system’s accessibility primitives after a refactor. The fix shipped inside a week because we had the data, the diagnosis, and a vocabulary to take it to engineering without translating.

Two things stuck beyond the save. The vocab stayed — design now writes quarterly narratives that the rest of the Exec team reads without a translator. And the dashboard stayed — it’s still running, owned by a rigor the team kept building on after I handed it off.

The work that doesn’t make slides tends to be the work that compounds.