2026
Master's final project (MSc UX Design)
Role
Product Designer
Timeline
8 months
Team
Product Owner, 2 UX Researchers
Research highlights
82% of survey participants couldn't explain where their energy was going
✶
Shifted the primary screen from dashboard to decision — a single Go/Wait recommendation, with supporting data one tap away
Moved from a standalone consumer app to a white-label model embedded inside the provider's apps
Validated the direction across two rounds of moderated testing, from lo-fi wireframes to the interactive hi-fi prototype
xxx
Context
In 2018, Spain became the first EU country to complete a nationwide smart-meter rollout, giving roughly 19 million households access to hourly electricity pricing under the PVPC tariff system — where prices can swing up to 4x between peak and off-peak hours.
In theory, that combination of universal smart-meter coverage and dynamic pricing gives every household everything it needs to manage energy costs.
The brief we set ourselves was simple: if people could see their consumption more clearly, they'd make better decisions and use less energy.
That assumption became the thing we spent the next eight months testing.
Discovery & research
We combined secondary research with 42 survey responses, 9 semi-structured interviews, and a competitive analysis of existing utility apps, monitoring tools, and behavioral-engagement platforms.
Rather than asking people what features they wanted, we asked how they currently made energy decisions — what they looked at, when, and why they stopped looking.
A consistent pattern emerged almost immediately: people already had access to consumption data through bills and utility apps. The problem wasn't visibility. It was that the data never turned into a decision they trusted.
xxx
Key findings
Data existed, but understanding didn't
82% of participants couldn't explain where their energy was going, and most only noticed a problem once the bill arrived — the information was technically available but never actionable in the moment.
Money motivated more than sustainability
91% said lower bills would change their habits, versus 50% for sustainability — financial framing, not energy units, was the strongest lever we had.
People wanted a decision, not a dashboard
The recurring question in interviews wasn't "what does this chart mean," it was "should I do something right now, and is it worth it."
Effort without feedback killed motivation
68% had already tried to reduce consumption but couldn't tell if it worked. Without a way to validate their actions, people quietly gave up.
Defining the problem
Problem
Households don't lack energy data — they lack a trusted, low-effort way to turn that data into a decision, and confidence that the decision made a difference.
Hypothesis
If people receive personalized recommendations that clearly communicate when to act, why to act, and the potential impact of those actions, they will be more likely to adopt and sustain energy-saving behaviors.
Direction
Design shifted from a monitoring tool to a decision-support product: recommendation first, supporting data second, savings shown in euros.
The pivot: from B2C to B2B2C
Our original concept was a standalone consumer app requiring smart plugs on every appliance — a real barrier to adoption before we'd even built the product. Competitive analysis raised a second issue: retailers already owned the customer relationship and the data, so acquiring users independently would be costly, with no clear way to charge for the service ourselves.
So we pivoted: Lumo became a white-label layer embedded inside retailer apps, inheriting their trust and distribution instead of competing with it.
One upside: without the hardware requirement, we researched non-intrusive load monitoring — identifying appliances from their electrical signatures, no smart plugs needed. We didn't build on it, but it shaped how we thought about future phases.
Key design decisions
The Go/Wait signal
Problem
Existing tools — PrecioLuz, retailer apps, smart-home platforms — all left interpretation to the user. None answered the one question people actually asked: should I act now?
Decision
A single recommendation became the primary screen: Go when prices are favorable, Wait when a cheaper window is approaching, with the reasoning always one tap away.
Why it mattered
It collapsed hourly pricing, consumption patterns, and tariff logic into one decision, removing the interpretation step every competitor still requires.
xxx
The provider-facing dashboard
Problem
Once Lumo shifted to a retailer-embedded model, providers needed a way to see whether the platform was working for their business, not just for households.
Decision
An analytics dashboard gave providers visibility into customer engagement, recommendation adoption, savings outcomes, and churn indicators.
Why it mattered
It gave retailers a concrete reason to adopt the platform — turning household-level behavior into business-level evidence of engagement and retention.
Financial framing over energy units
Problem
Kilowatt-hours and technical tariff terms didn't move anyone in interviews — money did.
Decision
Every recommendation, summary, and notification was rewritten in euros: "Save €0.40 this hour" instead of "reduce consumption by 12%."
Why it mattered
This aligned the product with the strongest motivator we found in research and made the value legible in under a second.
xxx
Progressive disclosure of supporting data
Problem
Early low-fidelity testing showed that even engaged users felt overwhelmed when charts and the recommendation competed for attention on the same screen.
Decision
Hourly pricing, historical trends, and neighbour comparisons moved to a secondary layer — always reachable, never required to complete the core task.
Why it mattered
Less engaged users could act in seconds, while data-oriented users could still dig deeper without added friction for everyone else.
Closing the feedback loop
Problem
Participants who tried to save energy had no way to confirm whether it worked, which was quietly killing motivation to keep going.
Decision
A weekly savings summary and progress indicator connected specific actions to specific euro outcomes.
Why it mattered
Turning invisible effort into a visible result was the biggest driver of continued engagement across both rounds of testing.
How we used AI
AI tools supported the process from the first sketches to the final prototype, always as an assist to our own judgment, not a replacement for it.
Research — helped synthesize interview notes, cluster affinity-map themes, and draft survey questions
Design critique — Claude ran heuristic evaluations, copy reviews, and accessibility checks at multiple stages, alongside human moderated testing
Screen and prototype creation — Claude and Figma AI helped generate early screen concepts and variations; v0 was used to build an initial interactive prototype
Every recommendation still went through our own research, testing, and decisions — AI sped up the process, it didn't make the calls
Validating the direction
Tested twice — 8 moderated sessions on the low/mid-fidelity prototype, then 5 on the interactive hi-fi prototype. The "use now vs. wait" idea was understood immediately by every participant. What needed fixing was tactical, not conceptual: in hi-fi testing, the pricing chart and a few controls weren't read as interactive, which fed directly into the next iteration.
The business model held up too — presenting the B2B2C pivot to professors with product and investment backgrounds meant it was challenged on acquisition cost and defensibility, not just design, and it survived.
One unprompted idea from the final presentation stuck with us: a widget showing savings and dynamic pricing inside open banking apps — extending Lumo's logic beyond the retailer relationship into where people already track their money.
Reflection
What made this project so valuable was that it took us through the entire design process — ideation, research, a pivot based on what we found, content strategy, business model, platform design, and testing.
Going in, I believed that giving people more visibility into their energy usage would naturally lead to better decisions. Research proved that assumption wrong, piece by piece.
The pivot wasn't comfortable — we had to redo a lot of our research and rework decisions we'd already made. But it taught me the value of that flexibility: recognizing when something isn't working and letting it go for something better, rather than pushing forward just because we'd already invested in it.
Beyond the design work, building Lumo as if it were a real startup — thinking through the business model, the roadmap, and what we'd need to show investors — made this feel like a product with a real future, not just an academic exercise.
