A high-stakes market entry, with no evidence behind the promise.

Tysons is Northern Virginia’s densest development corridor, and one of its most contested. Every developer there is making the same promise about walkability, amenities, and urban living. Gensler needed a data-driven strategy to create meaningful and impactful resident experiences in that market: something that would genuinely entice future residential growth rather than repeat the pitch everyone else was making.

Architectural and marketing decisions of this scale get made early and cost a fortune to reverse. The research had to land before the design approach locked.

24 focus groups. 150+ residents. One question, asked properly.

• Targeted recruiting: Screened and recruited across specific resident demographics living in and around Tysons: the people whose perceptions would determine whether the development succeeded.
• 24 moderated focus groups: Designed the discussion guide and led sessions probing perceptions and attitudes toward Tysons: what the place means to them, what works, what doesn’t, and what would actually move them.
• Cross-group synthesis: Analyzed 150+ participants’ input for the patterns that held across segments versus the ones that split them.
• Dual-track recommendations: Translated findings into two distinct deliverables: one for the architectural design approach, one for the marketing and positioning strategy.

The gap between how Tysons is sold and how it’s lived.

• Perception lagged the plan: What residents described about Tysons and what the development narrative claimed about it were not the same place. That gap was the strategic opening.
• Attributes split the segments: Residents did not weigh the same attributes the same way. Attitudes toward access, retail, green space, sense of place, and affordability separated cleanly by segment, which meant a single undifferentiated pitch would underperform with most of them.
• Experience beat amenity lists: What drew people in wasn’t the feature inventory. It was whether the place felt like somewhere you belonged. That is an architectural problem before it is a marketing problem.

Findings shaped the building and the campaign that fills it.

Gensler used the findings and recommendations to inform their architectural design approach for a multi-billion-dollar residential complex in Tysons, and to shape the marketing campaign used to attract new residents to it. The research moved from resident conversation to built environment.

Architects design buildings. Residents decide what they mean.

The most useful thing I brought to this wasn’t a method. It was the insistence that perception is data. A firm this good doesn’t need help designing. What it needed was a clear read on how people already there experience the place, before a billion dollars of concrete makes that experience permanent. Connection and community are design outcomes, not marketing language, which is why getting research into the room at the design-approach stage, not the marketing stage, is what made it count.

Political volatility. 19-day sprints. No room for waste.

CFPB is congressionally appropriated, creating a dynamic environment where goals shifted due to political pressures. Most projects ran on a 19 business day timeline, kickoff to report delivery. That meant meeting the team, identifying objectives, formulating a plan, executing research, and generating findings fast.

Two modes: exploratory when discovering, evaluative when validating.

• Exploratory: Focus groups for broader consumer behavior insights. Competitive analysis and environmental scans to evaluate existing artifacts.
• Evaluative: In-depth interviews (n=9) for existing assets: website, consumer tools, or prototypes. Eye tracking, heuristic evaluation, and IA optimization when possible.

Trusted partner. Consumer tools that improved financial literacy.

Created effective consumer-facing tools for financial literacy. Regularly presented findings to Directors and Department Heads. My team became a valued resource across CFPB.

Ad length was set by convention, not evidence.

Meta, Instagram, and X all monetize the same scarce resource — attention inside a feed built for scrolling past things. But ad presentation length was inherited convention. Nobody had tested whether the standard exposure was too short to register, too long to tolerate, or whether a running shoe and a financial product need the same number of seconds to land.

Getting this wrong is expensive in both directions. Too short and the impression never converts to memory. Too long and you burn feed real estate and user patience for nothing.

Eye-tracking every ad, on every platform, every round.

• Iterative rounds, not one study: Findings and recommendations from each round determined the ad types and presentation lengths tested in the next. The design converged on the answer instead of guessing at it up front.
• All three platforms, every time: Each ad was tested on Meta, Instagram, and X in the same round — so platform differences were a real comparison, not an artifact of testing different creative at different times.
• Multiple product categories: Tested across product types to determine whether ideal presentation time is a universal number or category-dependent.
• Two data streams, combined: Generated heatmaps and gaze plots to capture where attention actually went, then paired that with ad recognition and recall measures to capture what survived.

Looking at an ad and remembering it are not the same event.

• Attention has a ceiling: Recall did not climb with exposure. Past a point, additional seconds bought fixation without buying memory — the curve peaks and then flattens.
• Ideal length is category-dependent: Product categories did not share one optimal presentation time. Treating exposure as a single platform-wide setting was leaving money on the table.
• Same ad, different platform, different result: Identical creative performed differently across the three feeds. Feed context — density, pacing, what surrounds the ad — changed what the exposure was worth.
• Heatmaps alone would have misled us: Gaze data and recall data diverged. Ads that pulled heavy fixation didn’t always survive to recognition. Only the combined read gave a defensible recommendation.

250% ad revenue lift across platforms.

Recommendations informed changes to ad presentation across Meta, Instagram, and X, positively impacting ad revenue by 250% across platforms. The iterative structure is what made it credible — by the final round the recommendations were built on everything the earlier rounds had already ruled out.

The eye tracker tells you where people looked. It doesn’t tell you what stuck.

The temptation with eye tracking is to present a heatmap and let the picture do the arguing. Heatmaps are persuasive and frequently wrong on their own. Pairing gaze with recognition and recall is what turned this from a nice visual into a revenue decision. And running it iteratively — letting each round narrow the next — beat any single large study we could have designed at the start.

A portal used by every military branch — and nobody could figure it out.

The DCCP is the DoD's source for the DEOCS climate survey, administered annually to over 1 million service members across Marine Corps, Army, Air Force, and Navy. The UI was unintuitive, the information architecture was outdated, and most users didn't interact often enough to build muscle memory. Service administrators on 2-year rotations had zero structured knowledge transfer.

Stakeholder interviews. Heuristic eval. Then 60 usability sessions.

• Stakeholder interviews: Kicked off with leadership from DoD's Office of People Analytics and Service-level stakeholders.
• Heuristic evaluation (team of 6): Led 6 UX researchers evaluating the UI while completing user tasks. Findings informed the usability testing protocol.
• 60 usability interviews: Sessions with Commanders, Service Administrators, and Survey Administrators — each group received tasks aligned with their actual DCCP usage.

Three systemic problems. All fixable.

• Steep learning curve, no repetition: Users didn't interact often enough to build knowledge or muscle memory.
• Zero knowledge transfer: 2-year rotations with no structured way to pass tips from seasoned users to new ones.
• Unintuitive information architecture: Participants consistently described the information flow as outdated.

35% fewer hotline calls. Every year.

Recommendations were implemented — a more intuitive, accessible UI — resulting in 35% fewer annual calls to the DCCP hotline.

The scale made it real. The knowledge-transfer gap made it urgent.

The insight that stuck wasn't about the UI — it was about the 2-year rotation cycle. Every new admin started from zero. The design recommendations addressed the symptoms. The knowledge-transfer recommendation addressed the root cause. Both shipped.

Diverse stakeholders, constant turnover, and a site that needed a full rethink.

Stakeholders included active duty military, marketing, design, and research — each needing different language when we presented findings. Frequent duty station changes meant constantly onboarding new team members, which led to creating a research repository.

IA evaluation → brand identity A/B test → 14 rounds of sprint usability.

• Information architecture: Open card sort with Army recruiters, prospects, parents, and influencers. Validated findings with tree testing.
• Brand identity A/B test (n=150): Evaluated two distinct Army-branded Figma prototypes head to head to determine which visual identity users preferred — the result set the brand direction the rest of the redesign was built on.
• 14 rounds of sprint testing (n=9 each): Iterative usability testing as we implemented design changes round by round.

14% more recruiter contacts. Fewer wasted page visits.

Website analytics showed our redesigned pages improved navigation through fewer duplicative visits and a 14% increase in conversion rate for the "Contact a Recruiter" intake form.

Stakeholder interviews. Quantitative analysis. On-site field testing.

• Stakeholder interviews: With CFPB's Director of Consumer Response and CRC leadership.
• Quantitative analysis: Evaluated average complaint handle time using a representative sample.
• On-site QC assessment: Evaluated process workflow approach at each CRC location.
• Field eye tracking: Used Tobii X2-60 portable eye tracker to evaluate the triage UI Consumer Guides used.

$300K+ saved annually by reducing average handle time.

Multiple FDA Centers. Four audiences. Strict federal research constraints.

Each FDA Center (Tobacco Products, Human Foods, Biologics, etc.) had its own leadership and vision for the website. OMB guidance limited us to n=9 per audience segment without triggering a 4-6 month review delay. We had to devise a research plan inclusive of each Center's unique goals.

Analytics → heuristic eval → 4 rounds of audience-specific testing.

• Stakeholder interviews: FDA's Web and Digital Services team plus Center representatives.
• Quantitative analytics review: Identified traffic patterns to inform which pages we tested.
• Heuristic evaluation: Assessed adherence to design principles and uncovered usability issues.
• 4 rounds of usability testing: Each round focused on one primary audience segment.

ROI matrix, design workshop, and FDA-wide design guidelines.

Collaborated to generate an ROI/LOE recommendation matrix. Led an in-person design thinking workshop with FDA Directors. Culminated the project with an FDA Design Guidelines document to promote consistency across all divisions.

Stakeholder interviews → card sort → tree test → 5 rounds of eye-tracked usability.

• Stakeholder interviews: With NCI's CBIIT leadership team.
• Card sorting + tree testing: With bioinformaticists as participants to evaluate and optimize IA.
• 5 rounds of lab-based usability testing: 9 participants per round from targeted user populations, using Tobii X2-300 eye-tracking software.
• Task analysis + heuristic evaluation: Process workflows for primary and secondary tasks per audience.
• Personas + journey maps: Generated for each audience based on research insights.

The platform's purpose was unclear — even to NCI's own people.

• Identity problem: CBIIT's informatics role was unclear to internal researchers and extramural collaborators.
• Resource gaps: Key resources like ITCR and NCIP weren't accessible from the CBIIT platform.
• Poor IA: Information architecture was not intuitive.

Actionable recommendations for awareness, accessibility, and IA.