Case Study · Strategic UX Research · Fortune 500

What happens when a 100-year-old manufacturer discovers not everyone wants a smart home

Leading ecosystem-level research for Rheem × Harman International (Samsung), I uncovered that smart home adoption isn't blocked by technology — it's blocked by psychological diversity. The user who "just wants to shower" matters as much as the data enthusiast. This discovery reshaped the entire product strategy.

30%
UX improvement
post-redesign
4
Behavioral personas
redefining segmentation
3
Actor groups across
multi-device ecosystem
12 wk
Discovery to delivery
within consultancy sprint
Client
Rheem Manufacturing × Harman International (Samsung)
Role
Senior UX Researcher & Research Strategy Lead
Duration
12 weeks (May–Aug 2018)
Team
Cross-functional (offshore/onshore), Huemen Design consultancy
Rheem Smart Home Ecosystem
01
The Context

When a traditional manufacturer faces the IoT revolution

Rheem Manufacturing, a 100-year-old leader in water heating and HVAC systems, faced an existential challenge: transition from standalone appliances to connected home integration or risk losing market share to tech-native competitors like Nest and ecobee.

Their EcoNet system had the technical capability — water heaters and AC units could connect to smartphones — but adoption was stagnant, user confusion was high, and different teams across the organization held varying perspectives on what users actually needed from IoT connectivity. This wasn't a UI problem. It was a strategic alignment crisis masked as a usability issue.

Three actor groups, four device types, one unified system

The project's core complexity: this wasn't "user research" — it required mapping an entire business ecosystem where three distinct actor groups interacted with four different device types.

Homeowners

Direct users

Navigate purchase decisions, installation, daily usage, and troubleshooting. Pain points: intimidated by technology, uncertain about value, overwhelmed by data.

Contractors

70% purchase influence

Recommend products and bear responsibility for connectivity issues. Felt uncertain about their role with smart tech they didn't fully understand.

Technicians

Installers & repair

Required diagnostic data, error codes, and system specs — a completely different mental model from homeowners' consumer-facing needs.

Each group interacted with multiple devices — smartphone app (EcoNet), three water heater display form factors, AC unit controls, and physical thermostats — all requiring consistent terminology and interaction patterns across the entire ecosystem.

02
The Approach

Systematic hypothesis testing through provotyping

Rather than simply asking users what they wanted, I designed a research strategy that would systematically test internal stakeholder assumptions while uncovering latent psychological needs. Working within Harman's 24-hour design operation across time zones, I led the complete discovery-to-delivery cycle in 12 weeks.

Research architecture

  • Stakeholder Interviews — Surfaced internal perspectives, translated into testable hypotheses
  • End User Interviews (n=15) — Contractors/facility managers + smart neighborhood residents
  • Provotyping Sessions — Low-fidelity provocations testing data preferences
  • APP Usability Testing (22 surveys) — 60-minute sessions validating design directions
  • HMI Usability Testing — Physical device interface validation
  • Heuristic Evaluation — Nielsen principles assessment
  • Competitive Analysis — Nest, ecobee, emerging platforms

The provotyping methodology

Drawing from my doctoral research, I applied a provocation-based approach dividing interviews into two phases:

Phase 1: Understanding current behavior and pain points without bias.

Phase 2: Introducing low-fidelity provocations (paper prototypes without labels) to test stakeholder hypotheses.

This approach forced participants to focus on meaning rather than visual details, revealing psychological preferences that high-fidelity prototypes would have masked.

The information spectrum exploration

I created 10 different visualization sets exploring information across a spectrum — from device-centered (neutral status) to personal preferences to family dynamics to community insights — systematically testing which level of information density different users actually wanted.

Provotyping tool showing information spectrum
Provotyping tool: information spectrum from device-centered to personal to family to community levels — deliberately rough to force focus on meaning over aesthetics
03
The Discovery

Four behavioral personas: psychological diversity in smart home adoption

The research revealed something critical that internal stakeholders hadn't anticipated: users had fundamentally different psychological relationships with smart home technology. We couldn't design one interface for one "user" — we needed flexible architecture serving four distinct motivational profiles.

01

The Eco-Conscious

Values-driven · Community responsibility
"Want to receive education to help them make better lifestyle decisions"

Motivated by environmental impact beyond cost savings. "Being better in a sustainable way" was an intrinsic motivation driving self-learning and social interactions to share knowledge.

Design → Environmental impact visualizations, community features, transparent data sharing options
02

The Frugal Minded

Financial optimization · Data analysis
"Can make decisions on wattage and usage and unplugs things based on reports"

Comfortable with detailed graphs and numerical data. Wanted dollar amounts, not percentages. Willing to experiment — testing how low they could go, then checking the app to see financial impact.

Design → Detailed financial reports, prominent dollar values, predictive savings calculations
03

The Nonchalant

Simplicity · Minimal cognitive load
"I just want to shower." — Participant EU 8

"Feels overwhelmed by data." Satisfied with basic functionality. Not interested in reports — "Would look at it if there is something outrageous." This persona challenged the fundamental assumption driving the entire project: that everyone wants data optimization.

Design → Simple defaults that "just work." Alerts only for critical issues. Complexity hidden unless requested.
04

The Enthusiastic

Curiosity · Technological exploration
"Curious about ecological justice but unsure of the next steps"

"Liked that there were research opportunities." Wanted tutorials, system transparency, and the ability to understand how things work. Would try to repair the water heater themselves.

Design → Advanced features, in-app tutorials, reference guides, system transparency
Four behavioral personas with psychological profiles
Four behavioral personas mapped across psychological drivers, behaviors, and design implications — segmentation based on motivation, not demographics
04
Strategic Insights

Six discoveries that reshaped the product strategy

Insight 01

"Wasting" vs "Saving" — Loss aversion in action

"$5 saved doesn't feel like much. But $5 wasted? That feels like A LOT."

This single psychological reframe unified conflicting stakeholder perspectives. The product wasn't about "save money" (weak motivation) — it was about "avoid wasting money" (strong anxiety/regret prevention). This distinction drove messaging, feature prioritization, and data visualization strategy.

→ Reframed entire value proposition from gain-seeking to loss prevention
Insight 02

The experimentation pattern & interface trust differential

"Felt he would test the experience with the water heater to see how low he could go and then would check the app to see what the changes caused." — PAR 6

Users didn't want to be told optimal settings — they wanted to discover them through safe exploration. But critically, this confidence existed only in the app. With hardware screens, users were afraid because they couldn't return to initial settings. This app-vs-hardware trust differential had major implications for feature placement.

→ Designed for curiosity in app, safety cues on hardware
Insight 03

The contractor ecosystem gap

Contractors were critical gatekeepers — influencing 70%+ of purchase decisions — but felt abandoned. They lacked product education, worried about connectivity liability, and weren't comfortable pushing technology they didn't understand. This was a business model problem, not just a UX problem.

→ Identified need for contractor-specific features and education pathways
Insight 04

Privacy wasn't the concern — value was

Contrary to assumptions, participants were surprisingly open to sharing data. The real barrier wasn't privacy — it was uncertainty about value. When customers saw that contractors and service teams could provide better help with their data, data-sharing became a value proposition rather than a privacy sacrifice.

→ Reframed data sharing from privacy risk to service enhancement
Insight 05

Contextual learning: when users want education vs. reference

Users were open to education during installation and troubleshooting (high attention, problem-solving mode), but wanted quick reference during daily dashboard use. The opportunity: machine learning could surface specific savings at decision moments — "If you lower the temperature now, you'll save $12 this week."

→ Context-aware interfaces shifting between educational and reference modes
Insight 06

The autonomy-to-coaching progression (Self-Determination Theory)

The first condition for engagement was user autonomy — customers needed to feel in control. Technology that removed control was rejected. Only after autonomy was established would some users welcome coaching toward "being better." This validated SDT's emphasis on autonomy as a fundamental need, and revealed that prescriptive automation without established trust backfires.

→ Progressive pathways: autonomy first, then opt-in coaching
Collaborative research workspace showing synthesis
Research synthesis workspace: mapping findings across Self-Determination Theory frameworks and behavioral patterns
05
The Solution

Flexible architecture serving psychological diversity

Based on research insights, I led the design strategy for a unified information architecture serving four distinct psychological profiles across multiple devices and three actor groups.

Principle 01

Progressive disclosure based on persona

Default to simplicity (serving The Nonchalant), with clear pathways to complexity (serving The Enthusiastic and Frugal Minded). Users opt-in to data density rather than being overwhelmed by default.

Principle 02

Frame as "waste prevention" not "savings"

Alerts emphasized avoiding waste ("You're using 40% more energy than last month") rather than celebrating savings. Loss aversion drives action more effectively than gain anticipation.

Principle 03

Support experimentation with safe feedback

Rather than prescriptive recommendations, show consequences of choices — encouraging learning through exploration without judgment, especially in the app where users felt confident.

Principle 04

Unified terminology across devices

Three-tier system: Consumer Language (default), Technical Language (toggle for contractors), and Cross-Brand Recognition (acknowledging competitor terminology).

Information architecture: three-tier hierarchy

The design challenge was creating cognitive consistency — not just visual consistency — so users' learning transferred across every Rheem device they touched.

Information architecture diagram
Unified information architecture — three-tier hierarchy serving all actor groups across all device types
Before and after UI comparison
Before and after: simplified dashboard emphasizing clarity and progressive disclosure over data density
06
Impact & Reflection

What shipped, what shifted, and what I'd do differently

This was my first experience leading comprehensive research strategy for a Fortune 500 client within a high-velocity consultancy environment, and it fundamentally shaped how I approach complex product ecosystems.

Measurable outcomes

What I learned

Hypothesis-driven research creates velocity

Translating stakeholder perspectives into testable hypotheses via provotyping created immediate alignment. When people saw their assumptions validated or reframed with evidence, decision-making accelerated dramatically.

Psychological diversity beats demographics

Traditional segmentation (age, income, tech-savviness) failed to predict behavior. Psychological motivations — Eco-Conscious vs Nonchalant — were far more predictive of feature adoption and engagement.

Multi-actor research needs different methods

Researching contractors required different recruitment, different protocols, and different synthesis than homeowner research. The research architecture itself must reflect ecosystem complexity.

Constraints drive strategic focus

The 12-week timeline and 24-hour team operation forced ruthless prioritization. Every activity had to create business value, build reusable infrastructure, or challenge a critical assumption.

If I did this again

Methods & Tools

Research approaches

Provotyping Methodology Stakeholder Interviews Contextual Inquiry Semi-structured Interviews Hypothesis Testing Heuristic Evaluation APP Usability Testing HMI Usability Testing Competitive Analysis Persona Development

Theoretical frameworks

Self-Determination Theory Loss Aversion Psychology Research Through Provocation Progressive Disclosure Nielsen's Usability Principles Multi-Actor Ecosystem Mapping

Deliverables

4 Behavioral Personas Unified Information Architecture Visual Design System Stakeholder Analysis Wireframes (App + Hardware) Component Library Usability Testing Report Click-through Prototypes Visual Design Guidelines

Stakeholder management

Biweekly Client Presentations Cross-timezone Coordination Workshop Facilitation Executive Alignment Contractor Recruitment

Explore more work

See other projects demonstrating strategic research and systems thinking

← Back to All Projects
Next project → Why an analytics platform couldn't get anyone to talk