NoteDue to the sensitive nature of this work, some images are blurred or omitted — only publicly available visuals are shown.
The challenge
In fall 2022, home energy management systems were an emerging category and GM was entering it for the first time. There was no ecommerce presence, no established journey, and no agreed definition of who the customer even was.
The engagement began as a service design project with four goals: define how GM Energy should engage with and serve its customers; identify the systems, platforms, and processes needed to enable that experience; craft a future-state vision through concepts and prototypes; and develop the roadmap to bring it to life. The work wasn't redesigning an experience — it was deciding what the experience should be, then carrying it to launch and beyond.
Research
With nothing to audit, research had to build the ground truth. Desk research across GM's existing consumer and brand work, then fieldwork — GM dealerships to experience the electric offering firsthand, competitor EV dealerships, and generator dealers to learn how home backup power actually gets sold today. Consumer interviews and surveys followed, and a series of service design workshops kept GM stakeholders building alongside us: ecosystem mapping, cross-selling opportunities, archetypes, and collaborative journey mapping.
Mapping a journey that didn't exist yet
The research became an end-to-end journey map and service blueprint covering the full lifecycle — learn, shop, buy, install, own. For every customer step, the blueprint mapped touchpoints, frontstage staff actions, and the backstage machinery: which teams, which platforms, which support processes had to exist for that moment to work. For a company standing up commerce, an installer network, and financing simultaneously, the blueprint became the organization's shared source of truth for what had to be built — and it kept being revised for years as reality caught up with the vision.
Storyboards zoomed into the moments that mattered most — micro-views of a single step considering the customer, GM staff, and environment, across digital and physical worlds. The install consultation, for example: an authorized installer standing in a customer's garage, where the digital experience has to hand off gracefully to a person.
Personas that earned their keep
The first personas were built the usual way — stakeholder interviews plus secondary research — producing three EV-centric archetypes: Markus (electro-curious), Tom (Sierra EV buyer), and Fiona (other EV buyers). Dentsu's M1 segmentation sharpened the targeting: Tech Builders near-term, Smart Living long-term. A GM Financial journey workshop added Voice-of-Customer personas that pushed us to consider financing behavior explicitly.
The interesting part came later. Rather than let the personas calcify, we rebuilt them from evidence — pulling sentiments from every usability study run since May 2023, categorizing them by knowledge, cost, and concerns, then sorting real test participants into groups and deriving new attribute charts from what actual people said.
That analysis surfaced a segment the market was only beginning to recognize. Driven by increasingly frequent extreme weather, a group emerged that wants home energy security and has no interest in an electric vehicle at all — Rhonda, prioritizing HEMS over EVs. For a company rooted in vehicles, that's a strategic finding, not just a design one: the energy business has a market that isn't downstream of vehicle sales.
The personas became a working reference packet for the Home Energy Consultant team to facilitate sales calls and for the post-purchase care team — not a poster on a wall.
From vision to wireframes to launch
With the blueprint agreed, concepting turned into wireframes for the commerce experience — shop, PDP, bundles, cart, checkout, financing — then sprint-based delivery alongside GM's product and engineering teams. The launched experience sells the HEMS line — the GM Energy Home System, PowerShift Charger, and Vehicle-to-Home bundle — with configurable bundles, GM Financial financing inside the purchase flow, installation scheduling through certified partners, order tracking, and account management.
Designing for trust: financing inside checkout
The hardest problem wasn't the storefront — it was money. A HEMS purchase is a major financial decision, so GM Financial had to live inside the journey: a two-step prequalification where customers first see their buying power, then trade more information — including a full social security number — for real rates, terms, and monthly payments.
Asking for a social online is a trust cliff, and testing told us exactly how it felt from the other side. Participants said an unexplained SSN field makes a company seem "scammy"; that concrete rate, term, and monthly payment beat the abstraction of "buying power"; and that being asked to prequalify twice, without explanation, read as repetitive and eroded confidence — "I don't understand why I have to get prequalified again." The design response was education and transparency at each sensitive moment: why we're asking, what happens next, what you get in return, and what it does to your credit.
Underneath the visible flow sat combinatorial complexity most storefronts never face. Checkout wasn't one path but a matrix: hardware only, hardware plus installation, with or without a down payment, in states where financing was available and states where it wasn't, at full approval, partial approval, or decline — plus agent-assisted orders placed on the customer's behalf. Each permutation needed its own designed states and its own validation, and the team ran user-acceptance testing across roughly thirty distinct scenarios to prove the flow held together.
Building a testing practice, not just running tests
Post-launch, research became a standing operation. I helped structure it as two deliberate tracks so the team could both validate designs in flight and evaluate what shipped:
| Formal | Informal | |
|---|---|---|
| Intent | Retroactively evaluate shipped experiences | Proactively validate solutions mid-exploration |
| Scope | Comprehensive · fully built or hi-fi prototypes | Narrow · lower fidelity, one question at a time |
| Velocity | ~1 per month | 3–4 per month |
| Planning | A month ahead · ~4-week lifecycle | Coincides with the design release schedule |
Dozens of studies ran across both tracks — recruited through UserTesting.com, moderated and unmoderated, desktop and mobile, with quarterly check-ins to the client. A sample of what changed because of them:
- Forced sign-in removed from promo codesTesters found requiring an account to redeem a code unusual for a one-time purchase, and several read it as a privacy grab.
- Plain language beat specificationsA "What's Included" section outperformed technical product specs 7 of 10 as the thing that helped people understand what they were buying.
- Navigation simplified after a card sortGiven 24 energy products to group, most testers used four categories or fewer — far simpler than the taxonomy we'd built.
- Order status was effectively unfindableThe account dashboard card was almost universally overlooked; every participant went hunting elsewhere for their order.
We also acted as a critical partner to GM's own analytics team — reviewing their survey and heat-mapping studies and returning assessments and recommendations. One finding shaped the PDP directly: product specifications drew the highest engagement of any content (31%), yet only 45% of users ever scrolled far enough to see it.
"Overall, the feedback highlights a well-designed process with room for minor improvements in clarity, education and support."Post-launch usability readout — financing flow, N=16
Energy Genie: guided buying
A HEMS is a considered purchase — testers told us large purchases take weeks or months of research, that they prefer self-guided exploration, and that they're open to tools if those tools are accurate, unbiased, and genuinely personalized. That's the opening the Energy Genie was designed for: a guided experience that helps someone work out what they actually need.
For Genie 2.0 we tested two directions head-to-head — an open AI chat versus a structured survey — and the answer landed in between. Users relied on the system to lead rather than driving it themselves, and length didn't bother them as long as each step earned its place. The concept became a non-linear pathway merging AI dialog with survey questions in varying combinations, letting different customers enter and exit at different points.
Testing the shipped guided experience kept sharpening it: a vehicle-confirmation step was cut after half of testers found it redundant with the step before, and separate "Priorities" and "Goals" pages were merged after half found the distinction vague. The recommendations page itself rated 4 or 5 out of 5 with 80% of testers — the personalized reasoning and outage-risk section landed as "thoughtful" — but even the satisfied ones consistently wanted two or three options rather than a single answer, and 60% wanted installation detail before they'd feel ready to buy.
Designing the ownership phase
Selling the hardware is the easy half. In late 2023 the engagement expanded to the post-purchase experience with an explicit mandate: apply the same rigorous design approach to the ownership phase as the purchase phase.
That meant journey mapping and service blueprinting for everything after checkout — installation handoff, warranty, support, returns, escalation — across a partner ecosystem where GM owned the brand but not every touchpoint: installers, solar partners, logistics providers, and an EV Concierge contact center whose agents needed training plans built around the new journeys.
It also meant designing the feedback system itself. We mapped customer-satisfaction survey triggers across the entire ecosystem — post-sale, post-install, post-support-call, per partner — and then did the less obvious work of pruning them, prioritizing which moments genuinely warranted asking so customers weren't buried under surveys from four companies about the same purchase.
The wider system
As the experience matured, the work widened to everything around it: Home Energy Consultant booking flows, global navigation and sitemap architecture, CRM moment maps for the purchase, dealer-referral, and Genie journeys, agent-assisted ordering, error states, an executive dashboard for the business, and a roadmap sequencing ecommerce milestones against vehicle software releases. Accessibility ran throughout — WCAG audits of the live site plus a designer-facing checklist and Figma annotation kit covering contrast, keyboard operability, and motion, so compliance was something the team designed toward rather than remediated.
Quarterly competitive analysis across eight energy competitors — Panasonic, Schneider, Eaton, Enphase, Emporia, Generac, LG, and Span — tracked each one through awareness, consideration, purchase, and post-purchase, including their mobile apps. It confirmed the size of the bet: most competitors still have no direct ecommerce at all, only lead forms, dealer networks, and quote requests. GM Energy sells online.
Outcome
GM Energy went from no ecommerce presence to a launched, financed, installable product experience — and, just as importantly, a practice around it: a service blueprint the organization builds against, evidence-based personas that reshaped who the business thinks it serves, a two-track testing program feeding improvements into every release, an ownership experience designed with the same rigor as the sale, and a sales guide that puts the same journey in every seller's hands. Zero to launched, with the machinery to keep going.