Customer Service Evolved: Three Things From Octopus Energy’s AI Integration

The beginning of November saw us host our biggest Three Things Summit yet – ‘AI at Work: From Myths to Masterclasses’ – and kicking off proceedings was the visionary CEO and Founder of Octopus Energy, Greg Jackson. In his keynote address, Greg explained how Octopus has pioneered the integration of Generative AI into customer service to revolutionise an area of the business that was already functioning to an incredible standard.

Now the second largest energy provider in the UK, Octopus Energy set itself apart from its competitors predominantly through the quality of its customer service. “We really focused a lot on differentiating ourselves through service,” Greg shared, “and we work relentlessly to build the technology and culture to deliver that.” But that steadfast commitment to service was tested immensely when the energy crisis swept across Europe in 2022.

“Overnight, the wholesale price of gas in Europe rose fifteen-fold”, Greg recalled, “and so we suddenly had millions of customers facing bills they couldn’t pay. That meant double the calls, double the emails, and double the call duration that we were used to. We were really struggling to deliver the service we pride ourselves on, which was particularly painful in a time of such need.”

Embracing AI

Amidst such a trying situation, it might have been easy to ignore Generative AI’s entrance to the world stage, but ever the entrepreneur, Greg quickly realised how it could be a potent weapon for Octopus. “A friend of mine, who is particularly knowledgeable about that kind of thing, set up a WhatsApp group with ChatGPT integrated into it”, he revealed. “There were three of us in the chat, and for the next three days, we explored all the amazing things it might be capable of.” But, as it turned out, Greg wasn’t the only one at Octopus thinking about AI.

To his pleasant surprise, Greg discovered that his head of data science was already hard at work, figuring out how Octopus could harness ChatGPT’s power. “We had a big meeting with our global CEOs coming up, so I asked if he would be able to show us what he’d been doing. He demoed ChatGPT as a customer service assistant, and our jaws hit the floor.” Immediately, Octopus accelerated its efforts to implement AI to help with the deluge of customer queries it was faced with. “By January, I think about 5% of our customer emails were being answered with the help of AI”, Greg clarified. “By May, however, it was 45%, and now, it’s over 50%. That’s 50,000 emails per day, and about half of those are being answered in part, if not entirely, through generative AI.”

How Octopus’ AI Integration Evolved

Octopus Energy’s journey in embracing AI began with a simple, yet powerful application: using AI to draft replies to customer emails. “Agents would ask the AI to draft a response,” Greg explained, “and if they liked it, they would hit send; if they didn’t, they could simply edit it and then hit send.” This process developed over time, with the machine learning element of the AI picking up the writing style of each agent. If an agent liked to use emojis, for example, the AI started to use emojis, too, and the effect was incredible.

“The key metric that really emphasised the effectiveness of these responses came in February or March”, Greg revealed. “We saw that AI emails were getting 85% customer satisfaction ratings, whereas human-written emails were only getting 65%.” This proved that AI was not only helpful but had the potential to vastly improve Octopus’ customer service across the board.

The next leap in Octopus’ AI journey came from recognising that customer interactions are not isolated events, but part of an ongoing narrative. To create a response that truly satisfies a customer, the agent needs to understand the whole picture. Luckily, Octopus had been using a system, called Kraken, from the very beginning.

“Kraken records everything,” Greg explained. “It has recorded every phone call and transcribed it. It’s recorded every email, every click, every meter reading, every payment, even every failed payment; it’s got it all.”

Using the data collected by Kraken, Octopus could ask the AI to generate a summary of a customer’s entire journey with them. This innovative approach provided customer service agents with a comprehensive view of customers’ situations, enabling them to gauge how they would be feeling, and then craft well-informed and contextually rich responses capable of satisfying their needs. The success of this approach was so remarkable that it became standard practice for Octopus to auto-generate summaries for every interaction with every customer.

But Octopus didn’t stop there; one of the most useful applications of AI it has found so far is in dealing with complex cases. Greg explained, “It’s hard for humans to sift through extensive account histories to identify specific issues, like a misallocated payment.” By integrating traditional computing algorithms with AI, however, Octopus’ system now reviews a customer’s full statement, payment, meter reading, and communication history before formulating a response. “In these complex scenarios, our AI-driven approach consistently outperforms human efforts in delivering superior service,” Greg noted.

Currently, Octopus’ focus is on automating actions to streamline agents’ workflows even further with AI. “Consider a customer requesting a refund”, Greg suggested. “Previously, AI would determine the account’s status and draft a response, but a human would still need to manually process the refund. Now, however, while the AI is preparing the response, it’s simultaneously calculating the appropriate action to go with it.”

The upshot is that the only thing humans really need to do in this process is review the AI’s recommended response and actions, and authorise them. This advancement not only streamlines the process but significantly reduces the labour required, marking a new era in customer service efficiency.

***

Greg’s keynote was a fascinating way to begin our most successful summit yet. His account of Octopus’ AI journey provided us with so many insights and nuggets of advice, but below are three key takeaways that every business leader can benefit from.

Greg’s Three Things

1) Stay Curious

“I think that even in most start-up phases, the business kind of runs without the CEO”, Greg began. “Your job in that situation is really to spend time looking around the corner or working on the machine of the business to improve it, rather than being in the machine, yourself. And I think that exploration part of a CEO’s job is so important.”

So often we hear that businesses cannot stay still if they hope to survive, and with the advent of AI, that statement has never been more significant. As Octopus’ story demonstrates, the solutions AI can offer businesses are already exceptionally powerful, and they are only likely to become more so in the future. It is the curious leaders in the business community who will therefore benefit the most.

2) Store As Much Data as You Can

A critical reason why AI has been so successful at Octopus is the wealth of data it has to work with. If Octopus had not had so much data to hand, via their Kraken system, the effectiveness of their AI implementation would not have been nearly so pronounced.

“We were able to implement AI so quickly and so effectively because we had really good data”, Greg admitted. “Every time you don’t store your data, it’s lost forever, so it’s so important to do that. But another thing to remember is that if you store it in silos, it’s exponentially less powerful than if you store it holistically.”

3) You Don’t Have to Build it All Yourself

For leaders of smaller businesses, Greg’s experience, although interesting, might not seem that useful. After all, Octopus is a gargantuan company that can afford to develop its own software to integrate AI, so what if you don’t have that kind of budget?

“I look at some of the platforms that are available right now and I think they’re absolutely amazing”, Greg responded, “to the point that I have asked myself if we [Octopus] are engineering something we can just buy. I don’t think we are, but we have to ask that question because the solutions out there are so good. So I don’t think you have to be a big tech company or have a big tech department to get phenomenal benefits from AI, right now.”

***

We’d like to thank Greg immensely for providing our audience with such an honest and enlightening exposition of Octopus Energy’s AI implementation that we all learned so much from. Look out for more content from our latest summit in the coming days, as we continue to share the advice and insights our amazing guests had to offer.

If you would like to join us at our next Summit, apply to All Together today, where you can also access up to five hours of pro bono mentoring from a leading CEO, every year.