As a lead data analyst in Value-Based Care (VBC) program, I had spent a year building dashboards and reports – only to find hardly anyone was using them. Despite strong effort from our analytics team, adoption was low and stagnant. Users found the existing dashboards cumbersome, and many defaulted to old habits (like manual spreadsheets or ad-hoc requests). This personal frustration became the catalyst for action. I set out to uncover why our well-intentioned analytics weren’t sticking, and how we could turn things around. Through informal conversations, I learned that VBC leadership, clinicians and care managers valued the insights but were overwhelmed by the experience of accessing them. It became clear that the problem wasn’t the lack of insights or effort – it was the delivery. I pitched a bold idea: instead of more dashboards, we needed to treat analytics as a product and redesign the experience from the ground up.
Healthcare systems operate in a complex value-based care environment, managing quality and cost across a vast network. Data-informed decisions are crucial to succeed in VBC contracts. Teams faced systemic issues with data access and analytics. Care managers lacked an easy way to view patient metrics and trends in one place. Executive leadership had no clear view of how we are performing financially on at-risk contracts. Data analysts were inundated with ad-hoc data requests and had no centralized repository to share their work. As a result, valuable insights were underutilized and time was wasted duplicating efforts. The challenge was clear: how might we centralize analytics and make insights easily accessible.
Technical Product Manager (formerly Lead Data Analyst) – I initiated the effort and led it end-to-end, evolving from an individual contributor analyst into a cross-functional product leader. In practice, this meant:
In summary, I was responsible for driving the product strategy, guiding implementation, and delivering results. This involved not only technical skills but also stakeholder management with a common goal to embrace a new way of accessing data.
We kicked off with a design thinking workshop involving care managers, physician leads, and finance executives. We mapped out the Value Proposition Canvas for our two personas: what jobs they needed the product to do, their pains and gains. This exercise highlighted, for example, that care managers desperately wanted to “see all my patient metrics in one place” and avoid logging into five tools, while executives valued “a clear line of sight to cost drivers” without digging through data. These insights became our north star in design.
Unified Homepage with Personalization: A single-entry point where users could immediately see key metrics and navigate to detailed dashboards. We drew inspiration from Netflix’s interface – presenting content in an intuitive, engaging way. Instead of movies and shows, our “titles” were dashboards and reports. Categories like Quality Outcomes, Utilization, or Finance were displayed as horizontal carousels of dashboard thumbnails.
Screenshot above: The Value-Based Analytics (VBA) platform homepage. Users see a dashboard library with categories and thumbnail previews, enabling a Care Manager to visually pick a patient outcomes report or a CFO to select a finance dashboard as easily as browsing a streaming library. This design aimed to invite exploration and make analytics feel less like paperwork and more like an informative journey.
Global Filters System: To tackle the notorious filter fatigue, we built a global filter bar that persists across the platform. Users can set filters like date range, region, or population once, and those choices apply to all dashboards they view in that session. The Global Filters UI shows this persistent filter bar at the top of the screen. User could select one time and seamlessly switch between an outcomes dashboard and a cost dashboard without re-filtering. This not only saved time but also reduced errors and increased user delight.
Personalized Recommendations & Resume Flow: We wanted the platform to feel intelligent and personal. Our research showed that executive users often return to the same 2-3 dashboards monthly (e.g., a summary scorecard and a detailed cost drill-down). Meanwhile, care managers might not remember where they left off tracking a complex patient cohort. So we added a feature “Continue Where You Left” – the platform would surface recently viewed dashboards and suggest relevant ones based on the user’s role and viewing history. The Dashboard Recommendations UI below illustrates this. This predictive navigation meant users spent less time searching and more time acting on insights.
The platform’s recommendation panel highlights content tailored to the user. In this example, a regional finance leader sees a suggestion to revisit the “Quarterly Cost Trends” dashboard they viewed last week, and a new recommended report on “Readmission Hotspots” because others in similar roles found it useful. This personalization drove re-engagement and made the vast array of data feel curated for each user.
We developed wireframes and interactive prototypes of the homepage and global filters early. I conducted hallway usability tests with a few target users – e.g., asking a care manager to find a specific metric using the prototype – and used their feedback to refine the layout. One key tweak from testing was improving the visibility of the global filter panel (initially some users overlooked it; we enhanced its contrast and added a one-time tooltip “Set your filters here”). We kept the UI clean and focused – tight integration of Providence’s branding but with an emphasis on clarity and minimal clicks.
Our solution leveraged existing analytics infrastructure. We integrated with Tableau as the analytics engine for the dashboards (Providence already had rich Tableau dashboards for clinical and cost data). The platform we built was essentially a custom web front-end (“hub”) that embedded these Tableau views, managing navigation, filters, and user profiles. This approach meant we didn’t have to rebuild analytics from scratch – we added a user-friendly layer on top of it. Working closely with Data Analysts, we defined how global filters would translate to Tableau filter parameters behind the scenes, and ensured performance was optimized (caching commonly used data for faster load times, etc.).
Throughout the process, we stayed lean and focused: every two weeks, we reviewed progress with actual users or stakeholders. This not only de-risked the usability aspects but also continuously reaffirmed we were solving the right problems. Our north star was always the personas’ needs – if a feature didn’t clearly alleviate a Care Manager’s burden or empower an Exec to make a decision, it didn’t make the cut.
From the outset, we treated product development not as a series of static projects, but as a continuous discovery and delivery cycle, drawing on principles from Teresa Torres’s Continuous Discovery Habits, Marty Cagan’s Empowered, and Eric Ries’s Lean Startup. This meant we:
Rather than gathering user requirements once and vanishing into development, we ran frequent touchpoints with real users every 1–2 weeks. We scheduled brief feedback sessions—sometimes just 15-minute check-ins—to see how new features fit into workflows and whether leaders could easily get to the financial metrics they needed. We treated user interviews and prototype reviews as ongoing rituals rather than one-time events. This kept us close to the evolving realities of frontline work and ensured each iteration was grounded in authentic user insights, not assumptions.
We believed in releasing small, testable increments of functionality, measuring their impact, and iterating fast. For example, when we introduced the Netflix-style homepage, we limited it to just a handful of recommended dashboards and measured how often users clicked them versus navigating manually. Once metrics showed they preferred the recommendations, we quickly expanded that feature. Similarly, we tracked adoption metrics (like weekly active users and average session length) and correlated them with cost and quality outcomes to see if we were truly creating business value. Whenever something underperformed, we pivoted or refined it rather than pushing forward on a flawed assumption. This closed-loop feedback mechanism was crucial for turning user behavior into product direction.
We weren’t afraid to run small experiments that might fail fast. For instance, we initially tested an in-dashboard “note-taking” feature to let Care Managers annotate patient metrics. Usage data and user interviews confirmed it added complexity without clear value, so we removed it. This willingness to pivot or kill ideas that didn’t deliver results helped us maintain a clean, intuitive product and conserve developer time for features that truly served the personas.
In short, weaving together the best practices allowed us to stay tightly aligned with user needs, empower the team to solve problems creatively, and validate features in real time. The result was a living product that consistently delivered relevant value to Care Managers and Executive Leaders alike, ultimately contributing to stronger business outcomes and a more data-driven culture.
The initiative resulted in a dramatic turnaround in both usage and business value, far exceeding initial expectations.
Within 3 months of launch, active usage of the analytics platform jumped to ~90% of our target users, up from under 30% with the old scattered dashboards. This means nearly all intended users were regularly logging in to access data. Moreover, engagement was sustained: we saw 85% retention month-over-month, indicating that once users came on board, they kept coming back. Users reported that the unified portal saved them time and fit naturally into their workflow. One regional director said it became her team’s “morning ritual” to check the key metrics on the portal, something that never happened with the old tools.
Cost of Care Savings: An analysis showed an 8% reduction in cost-per-member in value-based contracts, compared to the previous year. While many factors drive cost, executives attributed a portion of this to increased visibility of cost drivers and waste – the product helped teams identify opportunities (like reducing unnecessary ER visits or high-cost claimants) and track interventions.
Efficiency & Time Savings: The Data Shop alone had a huge impact on productivity. In the six months after its release, we logged over 1,200 data downloads by users. Assuming each of those would have taken an analyst an average of 2–3 hours to fulfill (finding data, formatting, emailing), we estimate roughly 3,500 hours of analyst time saved annually. Those analysts can now focus on more complex analytics rather than repetitive data pulls. For the end-users, this meant getting the data they need in minutes instead of days – speeding up decision-making on the ground.
This initiative reinforced a core belief for me: product success is not just about delivering features — it's about solving the right problems, for the right people, at the right time. By staying grounded in user needs, aligning cross-functional teams, and iterating rapidly, we transformed analytics from something people avoided into a product they relied on.
As the VBA platform and Data Shop matured, adoption scaled across regions and stakeholder groups, we reached a point where value-based care contracts had quadrupled in volume — and with them, the complexity and scale of the underlying data exploded. The data engineering team faced new challenges: messy, unstandardized input feeds from disparate sources, growing expectations for real-time processing, and evolving definitions of performance metrics.
To solve this next bottleneck, I was asked to shift my focus upstream — to architect and lead the next chapter of our analytics transformation: building a scalable, modern data processing engine.
Curious how we built Hungry Wizard, the industry-leading tool that powers the analytics backbone behind all this?Click here to explore that story.
Or if you'd rather chat directly about anything from product strategy to analytics design or team scaling, I’d love to connect. Write to me here
