Centrica adopts Couchbase to give field engineers full view of customer data

Centrica, the British energy company and owner of brands that include British Gas and Hive, has deployed NoSQL database provider Couchbase in order to get a consolidated view of its line of business data so that field agents have a full view of the customer when out on visits.

The project was started following Centrica’s creation of a Hadoop data lake, which pulled more than 20 different field teams data streams into a central data pool. However, this set up still didn’t provide the company’s team with the insight that they needed.

I got the chance to sit down with Alan Fairhurst, head of integration delivery, digital technology services at Centrica, at this week’s Couchbase Connect Europe event in London, where he explained the project’s ambitions. Prior to implementing Couchbase, Centrica’s field engineers would end up in situations where, for example, the customer would have home service insurance, but the engineer would not have any idea that the same customer was also an energy customer. Fairhurst said:

British Gas has about 20 different field teams and each of those manage their own estate. All their own data is not visible to any other group, which makes it a challenge for field engineers that need to go into customers’ properties. The customer has an expectation that an engineer will have all the information about the customer. It makes for a very awkward conversation with the customer.

We set up a small team and we used Couchbase to expose a subset of the data that’s in the Hadoop data lake. The Hadoop data lake is pulling in all of the line of business systems.

We extract the information about the customer product holding. And also any information about the customer’s experience with us as well. Any interactions with us. So that the engineer can come in armed with the information. It’s not all the billing information, it’s not the highly sensitive information, it’s the key information for the engineer.

16 million records

NoSQL databases are becoming increasingly popular in a market where companies are faced with ever increasing amounts of data, and having to serve customers across multiple touchpoints. The traditional relational databases have proven unable to scale and provide the speed and reliability demanded by modern digital solutions.

I asked Fairhurst whether or not a relational tool was considered in the implementation. His response was conclusive. Fairhurst said:

It wouldn’t have given us the performance we needed. There are 16 million records. It needed to be very responsive from an app. The initial offering that we gave was an iOS app, which we called the Customer Information App. It needed to be very responsive, so when someone put in a postcode or customer details, they got a response back immediately. Couchbase was there for speed.

When asked why Centrica opted for Couchbase over one of the other NoSQL alternatives, Fairhurst said:

We wanted to go down the NoSQL route, partly because it was new and funky, but also because we did a beauty contest with others like MongoDB and Couchbase was the winner at the time. Couchbase was fast, reliable, and it offered the capability of managing everything in-house. There were sensitivities about the security, so it’s an on-premise solution. And we were given great support by Couchbase.

The future of Couchbase at Centrica

Originally Centrica’s use of Couchbase was dedicated to solve its standalone problem of customer insight for field engineers. However, since going live, more use cases are being found, more in line with Couchbase’s vision for becoming the engagement database of the enterprise. Fairhurst said:

Increasingly it is becoming a central part of our omni-channel vision. Once we had all that customer information in one spot, we are finding many more uses for it. For us initially it wasn’t an enablement database, but it’s evolving that way. At the moment we are using it more and more for engagement with our insurance partners, so we can quickly identify when a customer is a customer of ours. So we can offer discounts via the comparison sites. It’s also used increasingly for more dynamic pricing work, which is just an experiment at the moment.

However, Fairhurst did have a word of warning for other companies undergoing a similar project, where Couchbase could expand across the enterprise beyond a singular use case. This largely focuses on educating the enterprise on when Couchbase and its query language N1QL should be used. N1QL uses industry standard ANSI joins for manipulating JSON data, so SQL users can pick it up quickly. Fairhurst said:

We had an initial small team that was really focused on NoSQL and we really focused on that. It was quite a technical team, which were quite experienced. It was very close knit. We were able to learn about NoSQL quite quickly. When we broadened our offering and offered our use of Couchbase to other teams, there was a bit of a challenge and they abused N1QL.

The net result was that performance dropped off a cliff. I’d probably say that if I was to advise another company using it, make sure you get your training in place to really understand not only what you can get from Couchbase, but also how to use N1QL. When is the right time to use technology like Couchbase.