SystemsCarbon Direct2023

Helping non-finance teams get finance data

Carbon footprinting depends on spend data that sits with the finance team. I designed the guidance and upload experience that cut the back-and-forth between them.

My role: Product designer — research, journey mapping, interaction design, user testing
Focus: Product design, UX research, User testing, UI/UX

Carbon Direct's company operations dashboard listing emissions data sources, shown over a photograph of fog in a forest

Background

Carbon Direct’s measurement tool quantifies emissions across sources, tracks change over time, and surfaces areas for improvement. An organization’s footprint is calculated from two main data sources: operational spending and activity data.

The problem

Getting operating expense data is critical and difficult. The people doing carbon footprinting usually don’t sit in the finance function, so they depend on a finance team to pull it. The requirements almost always got messy — unclear data structure, format, or terminology — which produced rounds of back-and-forth and rework between two teams who don’t share a vocabulary.

How can we guide customers on what data to get, and where and how to get it?

Understanding the customer journey

I mapped the journey to locate exactly where the handoff between sustainability and finance broke down. The failure wasn’t in the upload step, which is where the product surfaced it — it was upstream, in the request the customer had to make of a colleague.

Measure user journey map with a persona panel on the left and five stages across the top: learning, onboarding, collecting and uploading data, reviewing and refining data, and receiving report. Rows below track goals, user actions, internal team touch points, feelings and thoughts, pain points, and opportunities. — Mapping internal team touch points alongside customer actions made the real cost visible — every pain point in the upload stage had a matching hour of our carbon accounting team's time.

Design iterations

Phase 1 — content and upload experience. Working from our own finance team’s feedback, I improved the guidance for exporting expense data out of accounting systems. The previous experience put the upload in a modal with a click-to-upload control; I moved it to a full page with a drop zone so customers could review the file after uploading rather than committing blind.

Phase 2 — pre-formatting the data. Our carbon accounting team had been manually cleaning and formatting every upload. We explored removing that step by auto-detecting data fields and doing the mapping automatically, then letting the customer review and edit the formatting themselves.

Phase 2 is the more interesting tradeoff: automating the mapping risks silently mis-assigning a category, which corrupts a footprint number a company may publish. Keeping the customer in a review step was slower by design.

Outcomes

I left the company shortly after release, so we never gathered the analytics. The metrics the team had planned to use:

Customers who share correct spend data without manual guidance from us
Time spent in the spend data upload feature before a first attempt
Number of uploads before the correct data landed
Volume of customer questions about collecting and sharing spend data

Naming the measures without getting to read them is its own lesson in shipping inside a startup — the instrumentation has to land before the designer does.