Simplifying complex workflows into research that clicks.

Direct Mail Optimization for a National Television Provider

UX research and prototyping to improve customer engagement and reduce costs.

The brief

A major national television service provider had a direct mail piece that consistently outperformed everything else in their mailing schedule, a so-called champion letter, but had no insight into what made it work or how to replicate that success. The ask was twofold: understand what made the champion letter effective at a specific, attributable level, and use those insights to inform a redesign that was cheaper to produce without sacrificing what made it work.

As the lead researcher and designer, I owned the full research lifecycle, from study design and recruitment through moderation, analysis, and synthesis, then transitioned directly into prototyping a redesigned version informed by the findings.

Mixed methods: why both mattered

We combined 25 moderated, in-person in-depth interviews with Tobii eye tracking in the same sessions. Self-reported interview data tells you what people think they noticed. Eye tracking tells you what they actually looked at. Running them together let us validate whether stated preferences aligned with actual behavior, and in several cases they did not, which was the most important finding of the study.

Eye tracking captured the critical first eight seconds of exposure to each of six mailers, the window most predictive of sustained engagement. I analyzed gaze patterns, fixation points, scan paths, and time-on-element data across headlines, imagery, and calls to action.

What the research revealed

01 — Simplicity outperformed complexity

The champion letter had the simplest visual structure of the six pieces: one clear CTA, a single benefit-focused headline, minimal clutter. Eye tracking confirmed it through more efficient gaze patterns and lower time-to-CTA.

02 — Visual hierarchy controlled gaze

The champion letter guided viewers along a predictable Z-pattern scan path: brand mark, headline, primary benefit, CTA, in sequence. Mailers with competing visual elements scattered attention and drove early drop-off.

03 — Trust cues were essential

Brand recognition elements, testimonials, and personalized addressing drove positive response. Mailers lacking these read as generic or overly promotional.

04 — Information overload actively hurt

Mailers with more than three CTAs, dense copy, or small fonts were the lowest performers. Participants skipped large text blocks entirely, regardless of informational value.

From research to prototype

I prototyped a redesigned one-page version in Axure: eliminated content eye tracking showed nobody read, restructured layout along the confirmed Z-pattern, retained every trust element, and reduced to a single CTA.

Reflection

The constraint of reducing cost without reducing impact forced the design toward greater simplicity, which turned out to be exactly what the research recommended anyway. The cost problem and the engagement problem had the same solution.

Discovery and Validation Research for a Legacy Platform Redesign

Research-led modernization of a B2B print ordering system.

A platform that had outlived itself

A major enterprise client relied on a B2B ordering platform in production since the early 2000s. Several key accounts had communicated directly that a modernized platform was a prerequisite for continued partnership. Redesigning on assumptions rather than evidence risked accelerating churn rather than reversing it.

I led the research program across the full redesign lifecycle, from initial discovery through pre-launch validation, and contributed to the design work itself across wireframes, flows, and interaction design.

Three phases across the lifecycle

Phase 01 — Discovery on the legacy system

Stakeholder interviews with client-side users, support staff, and account managers mapped workflows and conscious pain points. Observational sessions watched users complete real tasks in the live system, looking specifically for normalized friction.

Phase 02 — Iterative validation through design sprints

After key sprints covering ordering, dashboard, and account management, moderated usability sessions surfaced refinements: navigation labels that missed user mental models, form sequencing that confused, feature placements that made sense in isolation but not in real workflows.

Phase 03 — Final validation before launch

A comprehensive end-to-end validation served as a quality gate, confirming the cumulative design decisions held together and the discovery-phase pain points were actually resolved.

What discovery revealed

The ordering workflow was the highest-priority pain point, with users routinely backing out of partial orders and starting again, a behavior they described as normal. Mobile access was a requirement, not a preference. The dashboard was the most visited and least useful screen. And workarounds had become invisible: tabbed browsers as manual progress indicators, screenshots to track order details, paper notes compensating for missing system memory. None of it was reported as a problem because it had become part of the job.

Reflection

Users who had worked with the legacy system for years had genuinely stopped seeing its problems. Watching people work revealed issues no interview would have surfaced. Embedding research across all three phases meant the final product was not a guess.

Longitudinal Rapid Iterative Usability Testing for a Telematics Platform

A three-year dual-audience research program across multiple release cycles.

A program, not a study

A national insurance provider engaged our team to build and operate a sustained usability research program for their telematics platform, a connected suite of web and mobile products for usage-based insurance enrollment and ongoing participation. The platform served two distinct audiences: insurance agents facilitating enrollment, and consumers managing their own experience. I co-led the engagement with one research partner across all 14 sprint cycles over three years.

The sprint structure

Each sprint ran four rounds of testing in a six-week cycle: agents on the current interface, consumers on the current interface, a design and development iteration, then agent and consumer re-tests on the updated interface using the same task scenarios.

What the longitudinal structure made possible

01 — Dual-audience insight

Agents and consumers frequently had different mental models of the same interface. The enrollment confirmation page used language agents understood immediately because it matched their professional vocabulary, but consumers found it ambiguous and at times alarming.

02 — Tracking design decisions over time

By sprint 8, we could trace any current design decision back to its research origin. Each sprint documented issues by severity, design changes made in response, validation status of those changes, and newly introduced issues.

03 — Sprint-over-sprint quality improvement

Across 14 sprints, critical usability issues declined measurably. The product team began anticipating the kinds of issues we would surface, so later designs arrived more mature.

04 — Managing researcher fatigue

We rotated primary moderation across sprints, cross-checked each other's synthesis to prevent individual bias from calcifying, and documented our evolving assumptions explicitly.

Reflection

Testing the same workflows repeatedly across a product's evolution created institutional memory for the user experience: we knew what had been tried, what worked, what created new problems, and why. Findings that are not captured rigorously become unretrievable, and in a longitudinal program losing that thread has compounding costs.

Multi-Market In-Store Ethnography & Eye Tracking for a Heritage Apparel Brand

A mixed-methods retail research program across five U.S. markets.

Diagnosing the gap

A heritage apparel brand's core customer base was quietly lapsing and they couldn't pinpoint why. Their flagship jeans were still regarded as best-in-class and their stores still drew traffic, but sales were soft and competitors were pulling ahead in ways the client could feel but not articulate. I co-led a five-market program combining shop-alongs, wearable eye tracking, and focus groups across current, lapsed, and potential customers.

Five markets chosen to mirror the real customer

The brand's customer base skewed rural and middle American, so four of the five cities were rural or smaller-market locations, with one metro for contrast. In each market: two days of shop-alongs in client and competitor stores paired with in-store eye tracking, then two days of focus groups across the three customer segments.

What the research revealed

01 — The brand was losing touch with the everyday woman

The most consistent theme across focus groups was a sense the brand had stagnated while competitors moved forward. Customers still trusted the jeans, but felt other brands now offered comparable denim at better price points with assortments that felt more current. They weren't critiquing quality. They were describing a brand that had stopped evolving with them.

02 — Competitors were winning beyond denim

Competitor shop-alongs revealed more fashionable non-denim pieces at accessible price points. In markets where the customer had limited time and budget, the broader assortment was winning the trip.

03 — The jean wall worked, but was under-leveraged

Eye tracking confirmed the jean wall held attention, with consistent directional gaze patterns across markets. Certain zones received reliable attention while others were systematically under-noticed, giving the merchandising team a framework for placing high-priority products where customers actually look.

Reflection

Inside a competitor's store, lapsed customers articulated comparisons they would never have surfaced in an interview room. Concentrating the research in markets that mirrored the brand's actual footprint, rather than defaulting to easy major-metro recruiting, meant the findings reflected who was actually lapsing, not who was easiest to reach.

GOAT Builder

A mobile-first basketball game where you build a player from the ground up, then a simulation tells you where he lands in the all-time rankings. I designed, built, and launched the whole thing solo.

The idea

The endless GOAT debate is fun precisely because it never resolves. I wanted to turn that argument into something you could actually play with. You pick a position, draft signature traits off the legends, spend a fixed pool of points across ten attributes, choose an era, and then watch a career play out and get ranked against everyone who ever played. The whole loop had to feel satisfying on a phone in under two minutes.

Designing the build loop

The core design challenge was meaningful choice. Every attribute point and every stolen trait needed to visibly change the outcome, or the build would feel cosmetic. I tuned the system so that an extreme build, like a pure volume scorer with no defense, produces a genuinely different career and ranking than a balanced two-way build, which keeps people experimenting instead of finding one optimal recipe.

The simulation and ranking engine

Under the hood is an original simulation that generates a full career season by season, accounts for pace and scoring norms across eras going back to 1947, and then scores that career on counting stats, accolades, and team success to place it on an all-time ladder. Getting the ranking to feel fair was the hard part. An early version buried a thirty-thousand-point career outside the top seventy, which was obviously wrong, so I rebuilt the scoring weights and validated the fix by simulating thousands of careers and checking that the rankings held up against basketball common sense.

Shipping it

It is a single self-contained build deployed and live, with a share card so a result can be posted and challenged. Treating it as a real product, not a demo, meant sweating the things players actually feel: mobile layout, load speed, and a clean share preview.

Why it is in a research portfolio

My day job is research, but this is where I get to close the loop and build the thing. It shows product thinking, game and interaction design, and the discipline of validating a system against evidence rather than assumption, which is the same instinct that drives my research work.

Play GOAT Builder ↗