Innovation incubate goal
How might we create an engaging customer experience in a way that lowers rates?
Utilities are regulated so that profits are fixed. This means that in order for electricity to cost less per unit of energy- the utility needs to sell more electricity. And ideally it does this by encouraging customers to use more electric appliances while not spending money to upgrade costly grid infrastructure. They call this “Electrification”.
Customer interviews
Our team conducted 67 customer interviews over Zoom
Pain point interviews
We started with pain point interviews – asking open ended questions to understand the relationship between customers and their utility. How do they feel about how they use energy at home? What is it like to see their bills? What do they think about their utility?
Our findings:
It can be scary to open utility bills because energy use is hard to see while you’re using it and then you get all the charges at once at the end of the month. Because users can’t see the calculation on how their bill matches their usage, it leads to distrust in the utility.
Critically, PG&E is not part of the decision making process when customers are thinking about electrifying their home
Value proposition testing: concept cards and landing pages
We did 9 concept cards over 2 rounds. Between rounds we retired, adjusted, and added new cards. At the end of the interviews we conducted a spend test to find the most valuable features. We chose the top 3 concept cards and turned them into landing pages for PG&E programs which we tested again with new participants.
We summarized our results manually and compared them with the output from our AI tools. We found AI to be very good at summarizing interview feedback and a good tool for finding quotes later.
Our findings:
-Participants universally liked the idea of being able to see how their energy use was allocated between their appliances
-Themes of safety and comfort tested well. They also liked the idea that PG&E could notify them if something was wrong – be it appliance health or a high bill caused by overuse
-They didn’t want one-stop shop upgrades from PG&E – they didn’t trust PG&E would be neutral and thought other 3rd party contractors were already available to work with. PG&E was also notorious for being slow and hard to work with.
-Utility financing and 0% down loans were not novel or wanted
-They believed that programs offered by PG&E would be offset by rate increases
Prototype testing
I created 2 rounds of prototypes in Figma Make
We tested an app that showed participants what appliances were currently running in their home, how much they spent per appliance, and a variety of different electrification nudges (unusual usage alerts, age-based appliance troubleshooting, and EV offers in context with gas/electric simulations). Our product manager came up with several generous, but still possibly feasible electrification offers.
The prototype was customizable to reflect the participant’s actual home appliances and average bill in order to make the app feel more realistic.
Takeaways:
-
- Customers really don’t understand kW and need to see usage in DOLLARS
-
- We can’t just make a standalone app that explains energy use. It needs to include basic utility features like bill pay.
-
- Participants were skeptical of over reliance on AI in the app
-
- Participants were eager to share data about their appliances if they thought it would make their projections better and alert them for unusual usage or prequalify them for offers
-
- 9.5/10 would download
-
- 8.9 would recommend to a friend
-
- Usage screen was the big winner
-
- Offers were not particularly compelling
Quantitative study
Tested 3 different versions of the app – energy disaggregation only, basic utility features only, and combined
The goal of the quant study was to determine if we could build an app that only helped customers see their energy use or if we had to build a fully featured utility app. I built 3 different versions of static screens to show the app in the survey.
While we collected 1200 responses, data quality was really poor. When I looked into it, 40% of participants finished the study in less than 5 minutes and less than 25% of participants passed the knowledge check. I learned how to make pivot tables in Excel for this project and use Power Query to combine and clean data non destructively. However, after the data was clean, it wasn’t enough to be statistically significant. We ended up with 200 high quality surveys across the 3 arms combined.
Takeaways:
-
- All three apps scored similarly when rating the app, likeliness to download, and improvement in trust (but this was not statistically significant)
-
- There were clear trends in feature desirability order: high bill alerts was the winner followed by unusual usage alerts, projected bill, and spend to date
Desk research
We consolidated previous PG&E internal research, AI desk research competitive analysis, and interviews with several utilities
Dug up some stats on PG&E’s 5.6 million customers:
-
- 80% access PGE.com on mobile device
-
- Accessing PGE.com on a mobile device is growing year over year
-
- 73% are open to the idea of AI managing tasks for them within 3 years
Personas:
-
- Striving Manager (proactive / limited budget)
-
- Optimized Enthusiast (proactive / flexible budget)
-
- Overwhelmed Survivor (not proactive / limited budget)
-
- Comfort Seeker (not proactive / flexible budget)
I ran some competitive analysis by reviewing utility apps, app reviews, and 3rd party energy management apps.
I got to interview counterparts at Duke Energy, Alabama Power, ConEdison, and Georgia Power. They were able to share some numbers with us and anecdotally share how they built their mobile apps and lessons learned.
Solution & pitch
Create a scalable platform (mobile-first and AI agent-friendly) that allows PG&E customers to feel informed, empowered, and primed for electrification where customers view PG&E as a partner in achieving their goals
Our product managers ran numbers for estimated value of a mobile app to PG&E. We were able to get some real data from other utilities on their mobile apps and determined we could expect 11M in arrears reduction and call center improvement, 75M in demand response and Virtual Power Plant grid upgrade deferral, 50M in program targeting and electrification acceleration, and Priceless – CSAT score lift.
I contributed slides, user quotes, and the mobile app prototype for the final pitch presentation. The app used actual PG&E visuals and branding and proven information architecture from the Duke Energy redesign. The goal was to highlight a few critical use cases where a mobile app would shine for customers. It was not meant to be actual UI so it’s much cleaner and less detailed.
The pitch was accepted and the project moved into the acceleration phase.
Initial app explorations
Present home energy use in a clean, friendly interface
I had another week on my contract after the final pitch, so I helped the team with some initial explorations for data visualization. We asked AI to give some suggestions on views that customers may find helpful. I reviewed them with the team and we discussed what felt good, what was clearly ridiculous, and what we might want to develop further. Then I gave several views and variations to consider given my previous experience designing data visualization in dashboard products.
