01
Friction
For users
Households did not have the same equipment, goals or level of energy understanding.
Very different equipment
Strong savings motivation
Variable understanding
B2B2CWeb & MobileWhite labelflagship productE2E ownership
Personalize an energy dashboard
In 2018, our product helped households equipped with solar, batteries or electric vehicles monitor their energy. I replaced a single, generic dashboard with a personalized experience based on the household profile, to make the data more useful and the value more visible.
Challenge
My role
OwnedContributedDiscoveryProduct framingUX architectureInteraction designPrototypingUser testingDesign deliveryAnalytics reviewBusiness positioningWhite-label constraints
Year
Timeline
Tools
Understanding the problem
Methodology
User interviewsBehavioral analysisProfile segmentationComprehension testsEngagement analysis
Household energy profiles
34%
Consumption
only
27%
Solar
only
18%
Solar +
bat.
11%
Solar +
EV
10%
Solar +
bat. + EV
Households did not all have the same equipment level.
Bounce rate
48%Share of sessions leaving the dashboard without interacting with its key elements.
Time spent on dashboard
41 secAverage time spent on the dashboard per user session.
The single dashboard did not hold attention well enough.
Retention over time
100%
26%
9%
D0D7D30
Interest dropped sharply after the first discovery.
Motivation
Main motivation: savings
Households mostly consulted data that could explain concrete savings.
Interviews and engagement tests
How can we make the dashboard useful to different households primarily motivated by savings?
02
Friction
For the business
The product value became hard to demonstrate with a dashboard that was too generic or too technical.
Value barely visible
Limited engagement
Difficult differentiation
Exploration and Solution
01
Organize by need
Create dedicated sections: save money, understand consumption and track production.
User language
Goals more visible
Fragmented overview
Not very suited to hybrid profiles
Explored
02
Personalize by profile
Generate a dashboard adapted to the household's equipment, goals and level of understanding.
Immediate value
Needs better targeted
Sensitive configuration
Personalization to justify
Selected
03
Contextual progressive reveal
Keep one view, but progressively reveal the explanations and indicators that are useful.
Less costly
Improves understanding
Limited personalization
Problem less targeted
Explored
Check out the 3 key decisions I made
Personalization
Adapt the dashboard to the household
Because the same dashboard could not be useful to households with different equipment, motivations and levels of understanding.
Energy profile generated from onboarding, with motivations, equipment and level of understanding.
Avoided cost
- A view that was too generic
- Modules that were not very useful
Accepted cost
- A profile to explain
- Variants to maintain
Control
Let users adjust it
Because an initial profile could help at the start, but should not lock users into a frozen configuration.
Edit mode allowing users to change the dashboard widgets.
Avoided cost
- Frozen personalization
- Needs poorly covered
Accepted cost
- More interface states
- More complex logic
Pedagogy
Translate energy data
Because households were not looking for raw data, but for a clear explanation of their consumption and savings.
Transformation of raw energy data into understandable messages for the user.
Avoided cost
- Data that was not understood
- Value that was not visible enough
Accepted cost
- More UX writing
- Levels to manage
The impacts
Personalization improved the dashboard's key signals: initial attention, retention and perceived value.
The first visit became more relevant
with lower bounce and more time spent on the dashboard
Interest held better after discovery
with D7 and D30 retention increasing on personalized profiles
Value was better perceived
with savings-related modules consulted more often
Before
Bounce rate48%
Time on dashboard41 sec
Bounce rate36%
Time on dashboard54 sec
After
The dashboard held attention better from the first visit.
D7 retention
26%
34%
Before / After
More users came back after the first week.
D30 retention
09%
17%
Before / After
Interest held better after one month.
Module interactions
22%
34%
Before / After
Modules useful to the profile were consulted more often.
Retrospective
Design one view for everyone
DON'TAdapt the visible value to the user's context
INSTEADOrganize the interface around the product
DON'TOrganize it around what the user is trying to understand
INSTEADMeasure only global engagement
DON'TCompare signals by profile, module and usage moment
INSTEAD
