Insights

Writing on complex product design.

Articles on product design, UX architecture, design systems, and complex enterprise systems - organized by persona and topic, drawn from real projects.

13 articles

Shibolet & Co.

Engineering LeadersProject-Specific Data Visualization

Visualizing a Community, Not a Dashboard

Say data visualization to engineers and they picture analytics dashboards. Shibolet's internal community hub needed the opposite. It was a community hub for a law firm, not an admin reporting tool. The job was making events, courses, benefits, and interest groups scannable and inviting, with light cues like who's attending or active. No charting engine, no KPI grid. Project-specific data viz means the representation answers this domain's real question. Here it was always: what's happening in my community, and what's for me?

Read

Movement

Engineering LeadersProject-Specific Data Visualization

Visualizing Body Measurements and Results Without Pretending to Diagnose

Movement shows real body measurements and checkup results, to people who are checking in, not being treated. That boundary drove the engineering. The hard part of data viz isn't the chart, it's the data contract: reference ranges, units, result states, and graceful handling of pending or missing values across many orgs. Visualize for orientation, model for truth, and never let the chart claim more than the data honestly supports. Liability and clarity in one decision.

Read

Academic & Research Systems

Engineering LeadersProject-Specific Data Visualization

Visualization That Survives Real Scientific Data

Charting libraries assume tidy, modest datasets. Scientific data breaks that assumption hard — millions of points, irregular distributions, nulls that actually mean something. Generic visualization stacks either choke or quietly mislead. In research tooling I treat data viz as a joint design-and-engineering problem: virtualized or WebGL rendering for huge series, downsampling that preserves outliers instead of smoothing them away, progressive loading so structure appears first. And on integrity — honest axes, visible uncertainty, no hiding the points that don't fit. Expert users spot a misleading chart instantly, and then they stop trusting the whole product. Visualization here is bespoke by necessity.

Read

BlackSwan

Engineering LeadersProject-Specific Data Visualization

Visualizing Entity Relationships Without Drowning the User

BlackSwan asks opposite things of data viz: investigators need dense entity-relationship and ownership structure made legible; analysts need condensed triage they can clear fast. Same data, opposite goals. A naive node-graph dump is unreadable — the viz layer has to be opinionated about what to surface vs. defer to drill-down. For a CTO: these aren't bolt-on charts. They're purpose-built to the domain's data shapes. That specificity is what users trust under real caseloads.

Read

Zammit

Engineering LeadersProject-Specific Data Visualization

Visualizing a Shelving Configuration Without Requiring CAD

Zammit's flagship: a spatial shelving configurator for people with zero CAD knowledge. The visualization had to faithfully represent a manufacturable object while hiding everything CAD-scary. The engineering-interesting part: the picture is bound to the same constrained model the engine enforces, so what the user sees can never drift from what the factory can build. Configurator data viz isn't a chart on top — it's a real-time projection of the product model.

Read

Menora Insurance

Engineering LeadersProject-Specific Data Visualization

Same Data, Five Lenses: Visualization Under Multi-Axis Filtering

The hardest data-viz problem on Menora wasn't a chart. It was the agent-fee flow, where complexity comes from multi-axis filtering across many fields and conditional rules. Visualizing that means making the query itself legible: which axes are active, how they combine, what the result set actually represents. Then add five roles — agent, marketing, district manager, team lead, underwriter — each needing the same data through its own lens. For R&D: treat the view as a function of data + role + active filters, not a fixed layout. Faithful representation beats decoration every time.

Read

Shomrat Hazorea

Engineering LeadersProject-Specific Data Visualization

The Real-Time Results Feed Is Data Visualization Disguised as Shopping

It looks like shopping, but Shomrat HaZorea's Guided Selling feed is really data visualization: a live, filtered view of a large multi-attribute catalog that re-renders the instant a customer changes one criterion. For R&D it's the same problem as any responsive data view — coherent state, immediate updates, no stale or flickering entries. The feed has to visibly reconcile as choices narrow, because the user is trusting that what they see is what they chose. Treat the catalog as a live visualization, not a list, and the conversation starts to feel intelligent.

Read

Beehive

Engineering LeadersProject-Specific Data Visualization

Designing Beehive's Personal Zone: Visualization That Knows Its Job

Beehive's Personal Zone lets clients track activity and manage their info — it's a personal area, not a daily analytics power-tool. I designed the visualization to match that job: orient and reassure, not overwhelm. The engineering payoff of honest scope? Fewer charting dependencies, simpler data contracts, a maintainable frontend. CTOs: the cheapest, most reliable feature is the one you correctly decided not to over-build.

Read

Government & Public Sector

Engineering LeadersProject-Specific Data Visualization

Data Viz That Holds Up When the Public Is Reading the Numbers

A public-sector chart is read by citizens, journalists, and auditors — so a misleading axis isn't a styling nitpick, it's a credibility problem. Build charts as accessible primitives: truthful baselines, a tabular equivalent for screen readers, a headline figure anyone can grasp with drill-down for analysts. And keep them lean — they render on old hardware and slow connections. A chart that's only pixels excludes people and fails the audit.

Read

Ministry of Defense

Engineering LeadersProject-Specific Data Visualization

Visualizing Milestones the Public Can Trust

On the Ministry of Defense public portal, the moments a milestone or timeline gets visualized are the moments credibility is most at stake. Not a real-time dashboard — communication. So the engineering constraints led: accessible (not color-alone), legible on mobile, and graceful degradation so a failed chart never blanks out a government page. Clarity over flourish. On a public-record site, accuracy and accessibility ARE the features.

Read

Hacker U

Engineering LeadersProject-Specific Data Visualization

The Salary Table as a Behavioral Anchor

The most important data viz on HackerU's course pages isn't a chart — it's a potential-salary table, used as a behavioral anchor before you ever read the syllabus. The design decision that mattered to engineering: keep it structured data, not an exported graphic. A styled, component-driven table renders per course, stays legible on mobile, and scales across the catalog. Match the visualization to the cognitive job — and to what your team can actually maintain.

Read

Enterprise Mobile

Engineering LeadersProject-Specific Data Visualization

Designing Data Viz for a Phone in a Field Worker's Hand

Most data viz assumes a wide screen and a stable connection. Field enterprise apps have neither. The hardest part isn't the chart — it's showing a defensible truth on a small screen with data that synced hours ago over a weak signal. Mark data freshness honestly. Favor a few glanceable indicators over dense charts. Render from cache offline, reconcile on reconnect. For R&D: the viz layer and data layer have to be designed together. It's a decision tool, not a shrunk-down dashboard.

Read