Predicting Customer Satisfaction (CSAT) at the Household Level

Hello, and thank you for checking out the very first episode of Blast Point's new deep dive podcast. I'm Tomer Borenstein, the CTO and cofounder at Blast Point, and I will not be your host, at least not usually. Blast Point always strives to be on the cutting edge of AI and machine learning, and so our podcast is AI generated too. The insights, the data, that is all real and coming from experts at Blast Point. But the hosts, the music, that is all AI.

Welcome to the Blast Point deep dive podcast. We're thrilled to explore how data and AI are transforming customer experiences. This podcast, a collaboration between humans and AI, delivers expert insights to help businesses better understand and engage their customers. I'm Tom, your AI host, and I'm here with Anna, my fellow AI cohost, for this brand new AI generated podcast series. In each episode, we'll explore the power of data, predictive analytics, and AI driven solutions with practical examples and real world case studies.

Uh-huh. Yeah. I know that feeling.

It's like having a 6th sense. Right? And that's what we're exploring today, how to tap into that kind of predictive power, especially for you guys in the utility world.

Makes sense. Customer satisfaction is huge EE for utilities.

Absolutely. Yeah. So imagine being able to understand what makes your customers happy or unhappy before they even fully realize it themselves. That's the holy grail. Right?

For sure. But let's be real. Keeping utility customers happy, it's not exactly a walk in the park.

You got that right.

You've got service interruptions, billing issues that make your head spin, and those customer service calls don't even get me started. It's a lot to juggle.

The minefield out there. And that's exactly why this idea of predicting customer satisfaction is so intriguing. What if instead of just reacting to complaints, utilities could anticipate them?

Exactly. It's about getting proactive, you know, like a chess game, thinking a few moves ahead.

Love that analogy. But how do we actually do that? How do we go from reacting to predicting something as complex as human satisfaction, especially in an industry as unpredictable as utilities?

Well, the key lies in data and more specifically in the power of AI and machine learning.

Okay. AI and machine learning. Now I I gotta be honest. When I first heard about this, I was a bit skeptical. I mean, predicting something as fickle as human satisfaction using algorithms, it sounded, well, kinda impossible.

It does sound pretty futuristic. Right? But the truth is, this technology is being used right now, and it's already showing some impressive results.

Really? So we're not just talking theory here?

Nope. This is real world stuff. We're talking about predicting how satisfied a customer will be with their utility company, not just in general, but at the individual household level.

Hold on. Back up. Individual households. That's incredibly specific. It's like zooming in on a micro level. How is that even possible?

It all starts with data. Think of it like gathering ingredients for a really complex recipe. You're collecting all sorts of information, customer interactions, billing history, service usage patterns.

Okay. So we're talking about a mountain of data here. But what happens next? How do we go from raw data to actual predictions?

That's where the magic of AI comes in. It's like having a master chef step in, analyze all those ingredients, and somehow whip up a prediction, a satisfaction score for each household.

So the AI is like a master chef, and the data is our ingredients. I like that. But how does the AI actually make sense of all that data? What's the secret sauce, so to speak?

Well, it's not exactly a secret sauce, but it uses something called machine learning. Basically, it's a way for computers to learn from data without being explicitly programmed.

So instead of us telling the computer if this, then that, the computer is actually figuring out those patterns on its own.

You got it. And the more data it crunches, the smarter it gets, picking up on those subtle links between, say, someone's billing history and their overall satisfaction.

So let's say the AI looks at all this data for a specific household.

Uh-huh.

And then it spits out a prediction. How accurate are these predictions? Are we talking a rough guess or something more precise?

Well, the accuracy can vary, of course. It depends on the data itself, how much we have, and how reliable it is. But honestly, these models are often surprisingly accurate, even more so than those traditional satisfaction surveys.

Wow. That's pretty impressive. But here's another thing that comes to mind. What happens when things change? Like, let's say a customer has a really positive or negative experience after that initial prediction is made. Does the AI just stick with its first impression?

That's the cool part. These predictions, they're not static. They're constantly being updated in real time.

Wait. Real time updates. You mean, like, as new information comes in, the predictions adjust?

Exactly. Think of it as a living, breathing system. Always learning and refining its understanding of each customer. Say someone has a billing issue. Right? But it gets resolved quickly and efficiently. That positive interaction feeds right back into the model, and boom, their satisfaction score could get a little boost.

So it's like a continuous feedback loop, always taking in new data and adjusting accordingly.

You got it. And that's what makes it so effective for utilities because, as we've said, things can change in an instant.

This is all starting to sound a bit like mind reading, you know, but instead of a crystal ball, we're using algorithms.

It's a bit like that. Yeah. But here's the thing, predicting satisfaction is just the tip of the iceberg. We don't just wanna know what they'll feel, We wanna understand why.

You mean, like, why one household might be over the moon with their service while their neighbor is ready to jump ship to a different provider?

Precisely. And that's where things get even more interesting because knowing the why helps utilities take action.

It's like they say, the why is the way. So how do we crack that code?

Well, this is where we dive into the world of something called s h a p charts.

S h a p chart. Alright. You're speaking my language. Break it down for us. What are we talking about here?

So s h a p stands for Shapley Additive explanations. Sounds kinda technical. I know. Stick with me.

We can handle a little technical jargon. Lay it on us.

Think of a sh a chart like a decoder ring for customer satisfaction. It visually breaks down all those factors impacting the score for each household, positive and negative.

Okay. So it's like a personalized satisfaction report card for each household.

That's a great way to put it. The SHAP chart shows you which factors are really moving the needle on satisfaction. Is it their billing history, recent customer service interactions, maybe the number of outages in their area? The SHAP chart lays it all out clear as day.

So it's not just about knowing if someone's happy or unhappy, but really understanding what's driving those feelings.

Yeah. Exactly. And for someone like you interested in customer satisfaction, that's pure gold. Right? You can see what's working and what needs fixing.

I can see how that level of insight would be a game changer for utilities.

Alright.

It's like having a personalized playbook for each customer.

Mhmm.

But have any companies actually use this in the real world Mhmm. You know, with real results?

Absolutely. There's a fascinating case study, a large utility company. They supply power to a big chunk of the Midwest and the South. They were dealing with low engagement on their website, especially for those important customer service tools.

Yeah. We've all been there. Scrolling endlessly trying to find how to pay your bill or report an outage. Not exactly a smooth customer experience.

Exactly. They knew they had to make some changes, but instead of just guessing, they took a data driven approach. They used this AI technology we've been talking about to really understand their customers' online behavior.

So instead of a shot in the dark, they use data to guide their website redesign.

Precisely. And by analyzing that data, they pinpointed those pain points customers were experiencing on the site. For instance, they found a lot of customers struggle to find information about their bills or understand their energy usage.

Makes sense. Those are pretty crucial things. So what do they do with that information?

Well, they used it to strategically revamp their website. They made sure that information about billing, energy usage, all that crucial stuff, was front and center, easy to find, easy to understand.

So it's like they use the data to streamline the whole customer journey on their website.

Exactly. And the results speak for themselves. Within just 2 months, they saw a 30% jump in click through rates on their website.

Wow. 30%. That's huge. What does that translate to in practical terms?

It means customers found what they needed faster, easier without needing to call customer service or wait for an email response.

Which, as we all know, can be a major pain point. So happier customers and a more efficient system overall.

Exactly. And it's not just a one hit wonder. More and more utilities are adopting this AI approach to customer satisfaction with equally impressive outcomes.

It's amazing to think we're at a point where we can actually predict and shape something as complex as customer satisfaction using data. Yep. It's like we're entering a new era of customer experience.

I totally agree. And what's exciting is we're just scratching the surface. As AI evolves, I think we can expect even more personalized, innovative approaches to customer satisfaction.

That's definitely something to keep an eye on. So for our listeners out there, what's the key takeaway here? What can they start thinking about today?

It's simple. Data is power. Those who can harness the power of data to understand, anticipate, and meet customer needs, they're the ones who are gonna thrive.

Wise words to end on. Thank you so much for this incredible deep dive. For our listeners, take a look at the data you have. How can you use it to better understand and serve your customers? Until next time. Thanks for tuning in to the Blastpoint Deep Dive podcast. We hope you enjoyed learning how AI can predict and improve customer satisfaction. Don't forget to subscribe and share. Until next time.