Designing Operational Intelligence
Helping delivery leaders catch operational risk before it turned into delivery failure.
Approved by Leadership · Moved Into Build
A single platform replaced the manual interpretation of scattered operational data.
From Manual Interpretation to a Shared Read
The Problem Was Never Missing Data
The company already produced enormous amounts of operational data — project status, delivery metrics, quality reports, revenue — but it lived across different systems. The problem was never missing data. It was turning scattered data into a decision.
Some spreadsheets ran past 50 columns. Every Delivery Director built their own way of reading them, which made reporting inconsistent and decisions slow.
The platform wasn't built to replace spreadsheets. It was built to replace manual interpretation.
Two Kinds of User
Research showed people worked at two very different levels — and one interface couldn't serve both without overwhelming one group or oversimplifying for the other.
Design around the decision someone is making, not their job title.
Discovery Under Constraints
Reducing uncertainty, not collecting requirements
I couldn't shadow directors — these were live operations — spreadsheet access was limited, and most of the real knowledge lived only in people's heads. So discovery wasn't about collecting requirements. It was about reducing uncertainty before making product decisions.
Instead of asking "what should the dashboard show?", I asked what questions directors needed answered to run delivery well. Those questions became the product.
The questions that became the product
How Leaders Actually Decide
Discovery revealed leaders didn't simply monitor campaigns — they ran the same investigation every time something looked wrong. The product was built around that pattern.
The product wasn't designed to monitor work. It was designed to support decisions — mapping directly to the mental model leaders already used every time something looked wrong.
Defining Health, and Early Warning
One signal from four
Everyone talked about "campaign health," but no one defined it the same way — delivery progress to some, quality to others, revenue to finance.
Rather than pick a single number, health became a blend of four signals: how much work got done (throughput), quality, timeline, and available capacity. That combined signal drove a simple early-warning ladder, so problems showed up while there was still time to fix them.
Off Track is an early warning, not a failure. The real job was to buy leaders time between the two.
The early-warning ladder
Deciding What to Measure
Discovery surfaced dozens of possible metrics. The hard part wasn't collecting more — it was keeping only what changed a decision. Every candidate passed three tests.
That narrowed dozens down to six core measures. One definition reshaped the whole product: when is a task "complete"?
My UX instinct said after review approval. But the business planned around trainer throughput — even rejected work still consumed time and budget — so completion measured output, and quality became its own separate signal.
Understand why the business measures something before designing how it's shown.
Choosing the Bubble Chart
Why the obvious charts failed
The toughest visual problem: compare many campaigns across several dimensions at once, without burying the reader.
Tables were great for investigating but slow for comparing. Bar and pie charts showed too few dimensions. Scatter plots showed only two.
Chosen because it represented the problem best — not because it looked modern.
What the bubble chart encodes
Progressive Investigation
The real challenge came after spotting an Off Track campaign: getting to the exact pod or trainer responsible without making users restart their thinking. The obvious option — a new page for each level — reset the user's mental model on every click.
Instead, selecting a campaign didn't open a new page. The same table simply went deeper. Filters narrowed the data — throughput below expected, a specific pod, a delivery stage — and extra explanation like campaign names, metric definitions and chart periods appeared only on hover, keeping the screen calm.
Users should think about the problem, not about navigating the product.
Every Element Had to Earn Its Place
Enterprise dashboards tend to become a pile of charts and widgets. Here, Product and Design challenged every measure, chart and icon with one question: does this help a director make a decision? If not, it went.
Validate First
Doing less, on purpose
The long-term vision included recommendations and automation — but the first version deliberately did less. Its only job was to prove one thing: does clearer operational intelligence actually help directors decide better? Only after proving that would automation be worth building.
- Engineering confirmed the data actually existed before anything was designed — validate the data before the dashboard.
- Refresh ran hourly: enough to test the idea, without the cost of real-time.
Phasing
The Product in Action
A Delivery Director opens the platform on Monday with one question: which campaign needs me today?
Because size reflects business impact, the top priority is instantly visible. Selecting it drills the same table from Campaign to Pod to Trainer, narrowing to the root cause without ever losing context. Operational intelligence reduces the time it takes to decide where to look first.
What Success Looked Like
Success here wasn't screen time or visual polish — it was less effort to reach a decision. The questions that defined it:
Because this was a first version, the long-term proof — faster intervention, better delivery outcomes, fewer escalations — still needs real-world use to measure.
Leadership approved the direction, and the next phase was build, followed by roughly three months of use to evaluate.
User Interface
Selected screens from the first version.
Like how I work? Let's build the next one.
You've just read how I think — the research, the trade-offs, the decisions I defended and the ones I'd make differently. I'm open to Lead and Senior Product Designer roles, remote, hybrid or in-office. Tell me what you're building and I'll tell you where I'd start.