The Data & Insights Hub
Data was scattered, metrics often contradicted one another, and there was no clear path to action. I led Expedia’s performance experience redesign, first building a data foundation that 180,000 independent hoteliers could trust, then shipping the platform’s first GenAI coaching feature to turn that data into action.
The stakes
Hoteliers rely on Expedia's performance tools to understand their business: how they are performing, how they compare with competitors, and where to focus next. Most are independent hoteliers, without the dedicated resources or account support that chains enjoy, and they stand to gain the most when those tools actually work.
65%
of Expedia's partners are independent hoteliers
180,000+
independent hoteliers on Expedia's platform
$19M
of our team's $56M gross profit target
The problem
Independent hoteliers had no reliable way to understand their performance or act on it. Each session was a scavenger hunt across disconnected tools, and the data they found often contradicted what they had seen on other pages.
Scattered metrics
Hoteliers had to navigate multiple pages and reports to piece together a complete view of their performance. They struggled to find what they were looking for.
Inconsistent data
The same metrics, such as revenue, room nights, and conversion, showed different values in different reports. Hoteliers noticed and stopped trusting the data.
No path to action
Recommendations existed, but user research showed they often felt generic and self‑serving, rarely specific to a hotelier’s actual situation. Over time, hoteliers learned to skip past them entirely. The one feature designed to drive action had lost its credibility.
The approach
Solving this problem meant starting not with users, but with the organization. Two senior leaders held different, incompatible views on the future of the performance tools, each shaped by their own team's goals. Before any screens were designed, I ran an alignment workshop built around a cereal-box exercise: both leaders had to design the product's front-of-box pitch together. Working on the same artifact shifted the debate from organizational priorities to user value.
With that alignment in place, my design director and I then ran working sessions with market managers, Expedia's internal team that coaches hoteliers using the same interface. We had strong research on hoteliers but not on market managers, whose goals had recently changed. The key insight was that if the tool worked well for market managers, they would become advocates who could accelerate adoption among hoteliers. Designing for both was a distribution strategy, not just a coverage decision.
"My focus is getting hotels on member-only deals. That's what I'm being measured on."
— Jenn, Associate Market Manager
From there, I worked with my product manager, content designer, and researcher to define an 18‑month vision and an MVP that could be delivered within a single quarter. One deliberate cut was the advanced coaching layer, the set of features that would guide hoteliers toward specific actions. Getting the data right had to come first. The coaching layer launched in a second phase, once the foundation was in place.
The solution
Consolidating all metrics into a single view was partly a UX decision and partly a strategic one. Take length of stay: it could appear on several pages, each calculating it slightly differently. When those numbers were scattered, the gaps were easy to miss. Put everything in one place and the discrepancies become visible to both hoteliers and the teams responsible for the data.
One place for everything
A single dashboard brought all performance data together, organized the way each hotelier chooses. Metrics flagged as needing attention or representing an opportunity gave hoteliers a clear starting point from the moment they logged in.
Data you can trust
The metrics were owned by a separate team with different timelines and priorities. A series of demos to VP-level leadership made the gaps impossible to ignore, and the team agreed to a phased approach: launching with a limited set of metrics and expanding as each one was resolved. My content designer and I built a metric glossary defining how each metric works and how it is calculated. Those same definitions were surfaced directly in the product: hoteliers could click any metric card to understand exactly what they were looking at.
From data to decisions
Clicking into any flagged metric revealed not just what it meant, but what to do about it. ADD GOOD CONTENT
A zoom into AI coaching
Consolidation fixed the data problem. Knowing your numbers are right, though, does not tell you what to do about them. Hoteliers had seen recommendations before and learned to ignore them: they were too generic and rarely tied to their actual situation. This was also Expedia's first GenAI feature for hoteliers, and the design challenge was different from anything else on the project. A conventional UI is deterministic. An AI-powered one is not: every design decision had to account for a range of outputs we could shape but not fully control.
Before building anything, my content designer and I ran two workshops to define what good output looks like. The first was with market managers. We used it to understand how they present topics to hoteliers, in what order, with which proof points, and how they frame trade-offs. The second brought in ML engineers to define quality criteria and the guardrails for the outputs.
Generating insights takes time, and that became a design opportunity. Instead of a spinner, I designed a loading experience that walked hoteliers through what the feature was doing: "Analyzing your performance… Reviewing your market…" It wasn't a literal map of the technical process, but it was honest about the complexity and helped build trust before the output even appeared.
Even with careful design, AI outputs can't be fully validated before they reach users at scale. Some would inevitably miss the mark. We built a feedback mechanism in from day one: a thumbs up or down on each insight, with the option to share more context. The goal was to understand where the feature was adding real value, catch the cases where it got things wrong, and build the data needed to keep improving it.
The outcome
Senior leaders had their eye on one number: the recommendation completion rate, which directly links hotelier action to gross profit. It rose by 20%.
+20%
increase in recommendation completion rate
85.5%
of users in the test group stayed in the new version
A (83.1)
Top 10% of software products tested with the SUS method
Key takeaway
The hardest part wasn't the design. It was knowing what to ship first. Any experienced team can build a better dashboard. The real challenges were elsewhere: deciding what good enough meant for 180,000 businesses, knowing which post-launch signals to trust when the loudest ones pointed in the wrong direction, and building a foundation that two additional designers could extend without losing the thread. Hoteliers didn't need more data. They needed one place to trust it, and a team that trusted the direction long enough to get them there.