Xero, a global platform serving small businesses, has 3.95 million subscribers, many of whom are now spending less time on Xero Central, the company’s customer support and learning site. Nigel Piper, Xero’s executive general manager of customer services, considers this a positive development.
The company has observed a roughly 40% reduction in average customer search times and about 20% fewer searches requiring additional support. This improvement is attributed to the implementation of generative answers in Xero Central, facilitated by a partnership with Coveo. Customer feedback has been overwhelmingly positive, Piper notes.
“Our support has always been digital and centered around content, so integrating generative AI seemed like a natural progression,” Piper explains. “Our goal is to make finding information as simple and fast as possible. Since customers today are eager to embrace such technologies, this upgrade has been very effective. They are finding answers more quickly and leaving more satisfied.”
Xero Central’s generative answering system uses Coveo’s Relevance Generative Answering platform. This platform not only provides direct answers but also links to relevant articles and knowledge bases. Users get quick, accurate solutions, can filter and dive deeper into content, or even request personal support.
Xero has been working with Coveo for over five years. Initially, they integrated Coveo’s predictive search and recommendation engine into Xero Central. When exploring the potentials of generative AI, they naturally turned to Coveo again. Understanding the complexities and opportunities of generative AI required a trusted partner to provide insights, develop a framework, and ensure ethical implementation.
“For the generative AI project, we’ve closely collaborated with Coveo to understand its possibilities and safely harness its potential,” Piper says. “Generative AI offers a powerful user experience without replacing human expertise or altering our commitment to customer service—it enhances it.”
Customer service traditionally relies on extensive data and content. Making this information easily accessible and searchable is revolutionary. “Support areas will have to reinvent themselves because of this technology,” Piper states. “The amount of data we can now access to help customers is world-class.”
Xero has detailed records of nearly every question ever asked on its platform or through customer service. This data provides valuable insights into what customers are looking for, how they ask questions, and what answers are most satisfying.
Initially, Xero had to consolidate data for Xero Central, using a search tool to provide snippets of relevant information. Today, the search interface functions similarly to Google but with much greater depth. Users can pose questions in natural language, and the generative AI-powered search provides personalized, relevant answers and content.
For example, accountants often need different information than small business owners, and Coveo’s search tool can differentiate these needs. Generative AI allows content from various sources to be unified in a single view, as long as the data is properly mapped.
“That’s the incredible power of generative AI,” Piper says. “Organizations starting today don’t need exhaustive integration; they just need to reference their data sources. Relevant information can then be presented in a natural language response and accompanied by a list of resources.”
Personalization is highly valued by customers, who want their experiences to be tailored based on their data. However, ensuring data security is more important than ever. Responsible data use must accompany generative AI solutions to balance personalization with privacy.
“There’s a balance,” Piper says. “Respect what you know about customers while ensuring their data is secure. Make them feel seen and confident in their privacy.”
Monitoring how the platform delivers answers is essential. During development, data governance teams rigorously tested the platform. As it rolled out, expert evaluations ensured the accuracy and formation of answers. Feedback has been very positive so far.
“We wouldn’t be wise to assume this tool is perfect,” Piper adds. “We are committed to responsible data use and ensuring we deliver the right answers while protecting private data. The goal is always about quick and effective answers that meet customers’ needs and preferences.”
This article was produced by a company with a business relationship with VentureBeat.