Powerful data, buried under an experience people had learned to work around.
The tool helps a global marketing organization decide where to spend: which channels, which brands, which weeks. Analysts trusted the model. They did not trust the screens around it.
Edits were made one cell at a time across tables of hundreds of rows. Optimization results arrived as numbers with no story about what had moved or why. Metric names meant different things to different teams, and each of four markets had built its own workarounds.
- Audit the frictionReviewed the core workflows with analysts and SMEs, and turned what slowed them down into a prioritized list of UI improvements.
- Prototype with AIBuilt new features as working prototypes rather than mockups, so stakeholders could use an idea instead of imagining it.
- Test in four marketsRan stakeholder interviews and prototype sessions across the US, UK, Brazil and Turkey, and validated changes with SMEs before release.
- Keep one source of truthSet up a single space where the client reviews, engineers build and design iterates, with every requirement pinned to the work it describes.
Rule based bulk editing replaced cell by cell changes, turning the biggest blocker to scenario planning into a task of under a minute. Optimization results now explain themselves: what moved, where the budget went and how return changed. In product help and term definitions clarify metrics where they appear, and the shared prototype space shortened the loop from idea to approved build. The longer term work explores agentic AI to tailor the experience to analysts and executives.
