Presented by Coveo


Global small business platform Xero has 3.95 million subscribers, and a hefty chunk of them are spending less time on the company’s customer support and learning site, Xero Central, and that’s good news, says Nigel Piper, executive general manager — customer, at Xero.

It’s early days, but the company has seen the average customer search time decrease approximately 40%, while search sessions that require additional customer service support decreased by approximately 20%. It’s thanks to the rollout of generative answers in the Xero Central solution — a move aided by a partnership with Coveo. The response from customers has been overwhelmingly positive, Piper adds.

“Our support has always been digital and content-led, so gen AI felt like a natural way of progressing our platform — an organic extension of the way we already support customers,” Piper says. “The goal is always to make finding information as simple, intuitive and fast as possible. And because customers today are eager to embrace technology like this, we’re finding this upgrade is really hitting the mark. They’re finding the answers they want more quickly, and leaving more satisfied.”

Xero Central’s generative answering is built using Coveo’s Relevance Generative Answering platform, which delivers both generative answers as well as links to relevant articles and knowledge bases. Users get quick and accurate solutions to their queries, but can also filter and sort to take a deeper dive into the material, raise a support case or request a phone call from a customer support specialist for one-on-one assistance.

The power of industry-leading partnership

Xero has partnered with Coveo for more than half a decade, Piper says. They originally worked closely with the Coveo to integrate its predictive search and recommendation engine into Xero Central. Naturally, they turned to Coveo again when they were intent on exploring the opportunities generative AI is unlocking.

That said, the newly democratized iteration of gen AI that has seized attention across the corporate landscape contains a great number of question marks — from risks to rewards, best practices and true potential. A trusted partner was crucial in cracking the code, to not only offer insight into what it would take to leverage gen AI, but also the expertise to help develop a framework around how to think about the technology, how to deliver it, and what values and protocols are needed to set in place when your experiment goes live.

“For the generative AI project, we’ve worked very closely with Coveo to understand the possibilities of generative AI, and how we could safely and ethically capitalize on its potential,” Piper explains. “Gen AI gives our users a powerful experience, but it doesn’t replace human expertise, and it definitely doesn’t impact how we think about customer experience — it complements our goals.”

Customer service is ripe for disruption

Support for customers typically relies on a mountain of data, information and content, and the ability to make all of that information easy to access and easy to search and surface relevant information is a transformative breakthrough.

“Support areas will have to reinvent themselves in the future because of technology like this, because the amount of data that we can surface now to help customers better is absolutely world class,” Piper says.

For instance, Xero has a digital record of nearly every question that’s ever been asked from either the platform or customer service agents. That kind of data unlocks insight into what information people are asking for, what questions they’re asking, and which answers sent them away satisfied — as well as a record of the myriad ways customers frame their questions.

Originally, they had to map their data and bring it all into one place for Xero Central to access, and the search tool surfaced snippets, very much like Google, digging up the relevant sentence or the relevant path of the answer from the body of an article.

Today, the search interface is very similar to Google, but the dramatically increased depth behind that text box is remarkable. A user can ask a question using natural language, and generative AI-powered search and recommendation turns up the answer and content that’s most relevant to that specific person — Coveo’s secret sauce, Piper says. The platform knows how this question has been previously asked in the past and in what context as well as the answers that have been previously surfaced and bring that together with available customer data to provide truly personalized results — even more personalized if the user authenticates before searching.

For instance, accountants often need very different information than Xero’s small business customers, and Coveo’s search tool understands when a certain pool of knowledge is being called for. Plus, the beauty of gen AI is that content can reside in a broad array of places in a platform; as long as the data is mapped, the content can be brought together in a single view.

“That’s the incredible power of gen AI,” Piper says. “People and organizations starting today probably don’t have to go to that length. As long as they know where the data sources are and reference them, the relevant information can all be brought together for the user in a natural language response, as well as that list of resources.”

Data safety and overcoming risks

Personalization tends to be high on a customer’s wish list — they want you to use what you know about them, from who they are, where they are, and what products they use in order to make the experience more satisfying. However, keeping that data safe has become a more urgent priority than ever, so it’s crucial to adopt the principles of responsible data use along with generative AI solutions.

“There’s a sweet spot,” Piper says. “It’s about being respectful with what you know, while still being able to acknowledge the customer and their experiences and preferences, make them feel seen while also making them feel confident that their data is secure.”

It’s also critical to keep a sharp eye on how the platform is delivering answers to users. During the development stage, data governance teams worked behind the scenes to ensure the tool was locked down tight, thoroughly testing and re-testing. As they readied for release, they relied on a group of experts to evaluate the answers given, as well as how those answers were formed. Feedback continues to be very positive, and they’re confident in the usefulness of the tool so far.

“But you’d be a fool to suddenly say that we’re going to just assume this is the most perfect tool in the world,” Piper adds. “We have a commitment around responsible data use, and we’re doing the work to make sure that we’re delivering the right answers for customers in a way that also protects private data. That’s what it’s always been about. If this helps them get the answer quicker, or in a way that suits their style or their approach or their needs, absolutely fantastic.”


Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with TechForgePulse, and they’re always clearly marked. For more information, contact [email protected].