Concourse

I was given access to the benchmarking tool and familiarised myself with the types of data it could surface. I explored its search criteria, filtering and sorting options, and translated those capabilities into what they should look and feel like inside the Concourse platform.

I structured the experience around clear stages to make the integration legible as a sequence of decisions rather than a black box.

Connect and authenticate > Select dataset and engagement context > Import records and insights > Review cleaning and mapping outcomes > Apply formulas and benchmark logic > Resolve exceptions > Export the final file into the engagement.

Example design decisions: I gave analysts the option to define parsing parameters and preview the first 100 rows of data before committing to an import.

I gave analysts the option to apply formulas in-line within the same flow or reuse saved “recipes” from other engagements.

I introduced both a linear list of digital assets and a kanban‑style grid view of data files, so batches could be dragged into the next stage once processing was complete

Because research buy‑in was low, I used high‑fidelity prototypes as a concrete way for stakeholders to see and react to the flow before presenting the research plan. In the worst case, the team would still launch based on assumptions and treat any findings as a second‑round improvement; in the best case, I could test an interactive version with real users and feed those insights into the design before launch.

I involved the UK Research team, which was a first for this type of collaboration between the UK and US PwC teams. To address resourcing questions I framed it as an opportunity to test whether US teams would make greater use of UK research expertise in future product work.

I recruited participants via an email invite linking to a short Google Form, where they described their current workflow and selected preferred time slots for the session. I then worked with developers to spin up a tailored UAT environment, using my interview script and task list as the backbone for scenarios, and paired it with high‑fidelity prototypes that were reviewed and signed off by the PMs.

We aimed to understand whether analysts grasped the integration and its stages; had enough controls while querying the data; trusted the imported and transformed data; and felt confident enough to use the exported file in a live engagement.

The research validated both the value and learnability of the integration: analysts picked up the flow quickly and often began the next task before I’d finished the instructions. The interaction model felt intuitive to them, as they surfaced useful feedback on the new benchmarking flow and pointed to a few opportunities across the wider Concourse platform.

I synthesised findings in Miro (affinity clusters and a simple quadrant map) and presented them to senior stakeholders, prioritising recommendations into quick wins and broader experience themes.

• card UI patterns (ellipses for truncated titles vs three‑dot contextual menus)
• indecision about search defaults (design for browse-and-refine or precise-query)
• status visibility showing data batches being processed (too subtle in grid view)
• terminology and definitions such as “Ingest” for data and “Import” for files

The insights were well received and became an evidence base for UI refinements and cross-platform improvements.

We agreed to implement quick wins first and decided which issues to address before launch versus which to defer to a second iteration. The following sprint prioritised giving analysts a visual representation / UI of data mapping and the ability to edit each connection in real time.