My role
Director, Product Design
Sep 2023 — Current
My Team
4 designers · 2 researchers
Reporting to the CPO, I led Product Design and User Research for a completely new commodities pricing and market intelligence platform, spanning forecasting, cost modelling, dashboards and AI-powered tools.
2,000 new models
built by clients (in first 18 months)
120,000 models
viewed by clients (in first 18 months)
Usability improvements
over the legacy experience resulting in significantly less reliance on support
Cost Models was one of Expana’s most strategically important features* — here’s how we redesigned it to improve usability, support migration and drive commercial growth.
*
Customer data maturity at Expana
Discovery
Understanding the problem space & our legacy solution
Our design and research-led discovery spanned a range of internal and external stakeholders, including:
1
Our Customers, both existing and prospective cost model users
2
The Commercial team responsible for selling the cost models tool
3
The Customer Success team, who deal with user pain points daily
4
The Data team, subject-matter experts in pricing and forecasting
Output from an 'ask-the-experts' review of our legacy cost model workflow — surfacing strengths, weaknesses, and shared domain knowledge across teams.
Defining the scope of the redesign
The painpoints we found in early discovery pushed Cost Models towards the ‘radical overhaul’ end of our redesign spectrum* — calling for deeper discovery and a substantial design effort.
*
Redesign spectrum
Among the findings in our legacy experience were two important themes:
Theme #1
Confusing Experience
The experience was so complicated that our Customer Success team often built models on behalf of our users.
Theme #2
Difficulty finding data
Users struggled to find and select the right data for their models, getting lost in the breadth of Expana's dataset.
Addressing Theme #1 // Confusing Experience
Design for new users
I set the team a clear direction: a new user should be able to build a Cost Model without support, without compromising our ability to migrate existing customers.
Data entry
This needed to be intuitive enough to handle simple use cases while remaining robust enough for the complexity we knew was coming.
Watching customers build models in Excel showed us their familiarity with spreadsheets, informing our designs around editable tables.
We tested two different approaches for data entry:
1
The first design placed data input in modal windows, aiming to reduce complexity by letting users focus on one task at a time.
2
The second design used inline editing with live calculations in the table.
Users consistently showed a preference for inline editing - both power-users and less technically users.
The Excel-like keyboard navigation supported fast data entry, and was easily discovered in the tests.
To ensure users could distinguish between their own entered values and those calculated automatically by the system, we introduced cell styling: white for user-entered values, grey for system-calculated ones.
A simple visualisation above the inputs table helped users track their progress as they built their model, with clear messaging when something wasn’t right.
Addressing Theme #2 // Difficulty finding data
Helping users find the right data
Finding data was a significant barrier to building a cost model. Much of the underlying problem sat with data structure, which would take time to resolve properly.
To start, we identified some quick wins we could implement in the search tool:
1
We updated the empty state of our commodity search to show ‘recents’, to accommodate users who were often interested in series they had previously searched for or viewed.
2
We added a ‘series preview’ to the search tool, so that users could assess the suitability of the item (users could check if there were gaps in data, and see the performance of the item) before adding it to their model
Cost Model Outcomes
The redesign made Cost Models easier to use while strengthening their role in migration, retention and sales.
In the first 18 months:
2,000 new models built by clients
120,000 models viewed by clients
Improvement usability requiring less support
Ongoing improvement
Cost Models continued to evolve after launch, adding forecasting, better error handling, exports and customer contract data as we learned from real-world use.