Every utility talks about reliability. Almost none of them agree on what it means past the obvious answer, which is that the power stays on. Does it also mean a text before a storm hits? A bill that doesn’t swing without warning? A program you find out about before you need it? We had a hunch the definition was shifting, and that the shift mattered for how utilities talk to customers.
The hard part wasn’t the hunch. It was finding a fast, affordable way to test it before committing to a full research program built around a definition of reliability that might not hold up.
So we tried something newer, alongside our traditional research toolkit: a synthetic focus group, built and run by our own researchers with the help of AI. Two of them, one residential and one commercial. Here’s what we did, what it told us, and what we think it means for how utility marketers can put this tool to work.
Why Synthetic, and Why Now
Traditional qualitative research is still the gold standard for understanding real people. It’s also slow to stand up and expensive to run twice. Testing three message concepts, probing a hunch about fairness, and pressure-testing a discussion guide before you ever recruit a real panel takes real time and real budget, often before you know if you’re even asking the right questions.
Synthetic research closes that gap. At Mower, we call our version of this Mower Synthetic Intelligence: AI trained to simulate how a specific, real-world audience is likely to think, feel, and respond, guided the entire way by our own research team rather than left to run on its own. For this project, that meant building two full synthetic panels, ten residential energy decision-makers and ten small business energy decision-makers, weighted to match the actual nationwide mix of investor-owned, municipal, and cooperative utility customers reported by the EIA. Every respondent’s psychographic profile was grounded in third-party audience data, not invented from scratch.
Once the panels existed, our researchers ran them the way they’d run a real focus group. A discussion guide with warm-up questions, core questions, and probes. A moderator pushing for specifics instead of accepting vague answers. Independent first reactions to each message concept before any cross-talk, so no single respondent’s opinion could anchor the rest of the group, the same discipline we’d insist on in a live session.
What It Let Us Do
This research provided us with valuable insights much faster than a traditional study would have allowed at this stage.
We tested three different ways of describing a more proactive utility relationship, independently, in both audiences, before spending a dollar on production or media. One framing won in both groups, for the same stated reason each time. That’s a validated starting point for real message development, not a guess.
One advantage didn’t show up until after the initial sessions were technically done. A traditional focus group disbands the moment the session ends. Ours didn’t. Because the panel is a standing resource rather than a one-time booking, we could go back to the same ten residential respondents and the same ten business respondents days later and ask something new, without re-recruiting anyone or trying to get ten calendars to line up twice. When a specific angle turned out to be worth a deeper look, we simply reconvened the panel and probed it directly.
That changes how you think about a research budget. It’s not just faster. The panel is available on your timeline, not the field team’s, for as long as the engagement runs.
We started with a simple hypothesis: that reliability itself could become a broader utility brand position. The synthetic sessions helped us refine that idea. Reliability, it turns out, may function more as a baseline expectation, while predictability, proactivity, and honesty open up a more meaningful opportunity for how utilities communicate. A framing built around proactive communication resonated most strongly across both audiences, giving us a clear direction to dig into further. Keep an eye out for the companion piece, which explores those findings in depth. This piece is about the how and the why.
What It Won’t Do
We’d be doing utility marketers a disservice if we oversold this. Synthetic research is directional, not definitive. It’s built to generate hypotheses and explore likely reactions, not to replace a statistically validated study or tell you exactly how many customers will switch providers if you change a rate plan.
Ahead of building this offering, we ran synthetic versions of past real client studies side by side with the original fielded results. The pattern held up consistently. Synthetic panels are strong at replicating themes, decision logic, and segment-level differences. They’re weaker at predicting exact incidence, capturing the full messiness of real operational friction, or surfacing a genuinely unexpected finding nobody was already primed to look for. A synthetic respondent tends to know what a well-informed version of that person would know, which sometimes means it fails to be as confused, as inert, or as stubbornly loyal as an actual customer.
That’s not a flaw to hide. It’s a reason to be precise about when to use this tool, and when not to.
Where Synthetic Earns Its Keep
If you have a research budget and a list of open questions, here’s where we’d bring synthetic research in on your behalf, ahead of anything else:
Pressure-testing message concepts before you commit media dollars.
Give us two or three ways of framing a program or a reliability commitment, and we can tell you which one is likely to land, and why, in days rather than weeks.
Sharpening a discussion guide before it goes into the field.
Running a guide through a synthetic panel first tends to surface the ambiguous questions, the probes that don’t quite work, and the places where an assumption is doing more work than the data can support.
Exploring a new or hard-to-reach audience segment quickly.
Before investing in recruiting a niche audience for a full study, we can run a synthetic pass first to tell you whether the segment is different enough from your existing customer base to justify it.
Generating hypotheses to bring into real customer conversations.
A synthetic session gives us a strong first draft. It tells us what to ask real people next, and often sharpens the questions enough that the primary research does more with less.
What we wouldn’t recommend it for: measuring actual brand awareness, predicting a precise sales lift, or making a final creative decision on its own. Those still belong to traditional research, done with real people, at the appropriate sample size.
It’s a Complement, Not a Replacement
The strongest research strategy we’ve found treats synthetic and traditional research as stages of the same process, not competing options. Explore with synthetic. Validate with primary research. Measure with tracking and analytics once something is live.
Used that way, synthetic research doesn’t take work away from real human studies. It makes sure the studies you do run are asking sharper questions, aimed at the right audience, with a real hypothesis already in hand.
For an industry built on trust and long customer relationships, that discipline matters. Utility marketers don’t get many chances to test a message with real customers before it goes into the world.
Getting a head start on which questions are worth asking, before you spend the budget to ask them, is the advantage worth building into how research gets planned from here on out. We’ll be talking more about this approach, and about what the reliability research actually found, at E Source. Come find us there, or reach out to us here.