{props.sub}
: null}{props.intro}
: null}Turning subscription complexity into a clearer path from vehicle purchase to activation.
Getting owners to begin required more than simplifying sign-up. We had to make the service understandable, purchasable, and completable across a dealership, the web, and the mobile app.
The service gives more than two million owners access to remote commands, vehicle status, safety and security support, service tools, trip experiences, and subscription management. When I joined the account, usage was heavily concentrated around a few remote commands: remote start, lock and unlock, and vehicle location, despite a much broader connected-service ecosystem.
Senior Product Designer and design lead, working agency-side at Dentsu while embedded with Subaru of America. I owned product design across a system that involved mobile, web, retailer tools, vehicle interfaces, and multiple engineering teams: research, product strategy, information architecture, interaction design, visual systems, testing, retailer experience, and implementation.
That model became an important constraint: a design improvement only mattered if it increased customer value without creating disproportionate operating cost. It is also why several obvious ideas, such as refreshing vehicle data more often, never made it past a first conversation.
{s.p}
Reading the revenue composition, this case works three levers inside the profitability model:{" "} attachment rate at point of sale,{" "} enrollment start rate, and{" "} enrollment completion.
All three are set during onboarding. None of them are set by a single screen.
{x.p}
How might we reduce onboarding friction and increase subscription conversion through{" "} clearer value, a{" "} stronger point-of-sale experience, and a{" "} consistent activation journey?
{row.d}
Among owners who do not use the app, 75% do not know, understand, or recognize its value.
Owners generally did not understand what they were paying for, if they were paying at all. The product was capable. The value was invisible.
The survey behind that finding asked non-users to name their reasons. Ranking them mattered less than noticing that the seven answers were not seven independent problems.
Read as a list, this is a backlog of seven unrelated fixes. Read as a diagnosis, it collapses into three kinds of breakdown, each of which calls for a completely different intervention.
{f.p}
This became more than a research summary. It gave product, design, retail, and engineering a shared vocabulary for diagnosing where growth was breaking down, so a disagreement about a feature could be re-stated as a disagreement about which friction it was supposed to remove.
The reading rule is simple and it is what made the matrix useful in a room: an owner who had never heard of the advanced tier has an information problem, someone who understood the tier but chose not to begin has a value problem, and someone who started enrollment but abandoned it has a journey problem.
I combined the share of each owner segment, blocker severity, and a behavioral weight informed by the Theory of Planned Behavior. The weights were judgment, not measured causal coefficients, so I tested the ranking against multiple weighting combinations. The top priorities remained stable.
The model was never intended to make the decision for us. Its value was giving teams with different incentives something concrete to evaluate together, instead of relying on "I think users probably..." Retail, product, and engineering could argue about a weight, which is a far more productive argument than arguing about taste.
{x.p}
I deliberately addressed pricing clarity before point-of-sale design. The checkout experience could not explain a product that the organization itself described inconsistently, so building the transaction first would have shipped the confusion faster rather than removing it.
The journey map below traces the same enrollment path end to end, with the checkpoints marked where the friction actually lives. Selecting a checkpoint jumps to the design work it produced.
The first audience for pricing clarity was not the customer. It was Subaru itself. Following the rebrand, Companion Plus incorporated the features of Companion, while enrollment, upgrade, downgrade, cancellation, and refunds were distributed across different administrative flows. Five teams could describe the same package in different ways.
The original retailer interface is live further on, in Iterations: working through it is the fastest way to feel why a visual fix would not have been enough.
Instead of making the billing hierarchy easier to read, I proposed changing what the interface represented. Each tier became a selectable object with its own plan, price, term, status, and included services. Dependencies still existed, but the system enforced them rather than asking a retailer to remember them.
The retailer is mid-conversation with a customer and needs to change a subscription without second-guessing what a button will do. Status, term, and next bill are surfaced on the object itself so the state of the account is readable at a glance.
I carried the same object model into the customer experience but changed the information hierarchy around the customer's job. The customer-facing version adds progressive disclosure, included-service details, pricing, renewal information, and payment-term choices.
Owners who enrolled at the dealership could download the MySubaru app, but account creation still pushed them onto the web. That hand-off introduced friction at exactly the moment when purchase intent was highest: standing in the showroom, phone already in hand.
I added mobile activation and took ownership of the experience layer across eight authentication and account states already defined by engineering. Instead of one long form, the experience uses shorter sequential steps, clearer progress, and a direct path from dealership setup into an activated account.
Research showed that 71% of owners were most open to learning about connected services during vehicle purchase. That gave us a high-value moment, but two distinct problems to solve inside it: the customer needs time and independence to understand the package, and the retailer needs visibility into whether the transaction is progressing successfully.
I designed a contactless self-checkout flow in which the retailer generates a QR code and the customer explores the packages and completes payment on their own phone. Independence solves the customer's problem and creates the retailer's: a person who has handed the transaction away can no longer see it.
Failure states carry enough context for the retailer to assist without restarting the transaction. The retailer is not a bystander in this flow. They are the support layer, and the console is designed to tell them when to step in and when to stay out of the way.
Checkout solved transaction friction, but it did not solve the deeper value problem the research had surfaced. A smoother way to buy something you do not understand is still a way to not buy it.
I proposed an interactive showroom kiosk where customers could experience MySubaru before enrolling: trigger simulated remote commands, explore connected-service features, understand which capabilities belong to each tier, and see the value of the mobile experience in context. I designed the concept with our UI designers and copywriter using the redesign work running in parallel, so the showroom experience represented where the product was going rather than merely documenting the legacy app.
The comparison itself was deliberately narrow: the same fields, the same validation rules, the same copy. Only the segmentation and the progress feedback changed, so a preference could only be about structure.
A device without a passcode cannot safely support biometric login or stay-signed-in behavior. Engineering's proposed solution was technically correct: label the phone an "unsecured device" and hide the unavailable features.
I disagreed for two reasons. The message described a technical state without giving the owner a path forward. And hiding Face ID removed any opportunity for users to discover that the capability existed at all, which quietly costs a security feature its adoption.
I worked with our copywriter on four messaging alternatives, ranging from the original technical wording to an explanation that connected the required action to the benefit it unlocks.
Two interaction treatments: hide the feature entirely, or keep it visible and explain the condition when the user reaches for it.
We shipped the tested version. The important part of this decision was not winning an argument through stronger design taste. It was making my opinion falsifiable and giving engineering and product evidence they could evaluate with me: if 60% had chosen their wording, we would have shipped theirs.
{r.p}
The customer end of the same model, before and after, both live. Each one walks its own subscription screen; hover either to take over. The two builds are what the pricing-clarity argument looks like from the side of the person being asked to pay.
The portal prototype is a faithful interactive rebuild of the redesigned screens, not production software: plan data is fixed sample data for one account, and the retailer actions confirm rather than call a billing service. The selection behavior and the dependency enforcement, which are the parts under evaluation, are real.
Complete onboarding without leaving mobile. This flow closes the gap between dealership enrollment and account activation by letting customers create and link their account directly inside MySubaru, instead of being emailed to a desktop browser at the moment their intent is highest.
The sequenced structure was tested and translated into implementation stories by our PMs, and the passcode and Face ID messaging inside this flow shipped in the tested wording. I owned the experience layer across the eight authentication and account states; the states themselves were defined by engineering before I joined the flow.
One transaction, two coordinated experiences. The customer can explore and pay independently on their own phone while the retailer sees enough progress to intervene when support is actually needed. Both sides are live below, and they are designed to be read together: the console exists because the phone takes the transaction out of the retailer's hands.
Research put 71% of owners at their most receptive during vehicle purchase, but receptive is not the same as unhurried. Handing the transaction to the customer's own device gives them the one thing a shared counter screen cannot: time to read without a person waiting on the other side of it.
The customer reviews all three plans at their own pace, expanding and collapsing what each one includes, and confirms payment without a retailer reading the plan sheet aloud to them. This is the same tier content as the app, on the surface where the purchase decision is actually being made.
Demonstrating value before enrollment. The kiosk lets a prospective connected-service customer explore real MySubaru behaviors in the showroom instead of relying on a verbal sales explanation: trigger simulated remote commands, see which capabilities belong to which tier, and watch the app respond while their vehicle is being prepared.
The tour advances on its own so the screen is never dead in a showroom, and the board stays fully interactive if a customer takes over. The app screens it shows come from the redesign running in parallel, so the demonstration represents where the product was going rather than documenting the legacy app.
Three dealerships, with success criteria agreed before launch and both in-flow and in-person feedback collected. Those results determine whether the concept scales toward Subaru's roughly 600-store retail network. Nothing here was rolled out nationally, and the pilot was scoped specifically so that a negative result would be cheap.
All-level MySubaru enrollment.
Adoption of the paid, value-generating tier.
{x.p}
These are business outcomes associated with the broader onboarding and activation program, measured across a staged rollout rather than a controlled holdout. The onboarding work shipped alongside other product, retail, and marketing changes, so the honest claim is contribution to an environment, not causation by any single interface. Where a specific design decision does have a controlled result behind it, such as the 61.3% and 60% preference findings, that result is reported in Testing with its method attached.
Onboarding solved the first growth problem: helping owners understand the service, enroll, and activate it. The next question was harder: what would make them come back when they did not need a remote command?
That question became the MySubaru Experience Redesign, the retention half of the same program.