Xero Incorporates Machine Learning for Bank Reconciliation

Xero, the global small business platform, today announced a significant milestone in its AI strategy with the roll-out of new bank reconciliation predictions. The predictions feature brings a new application of AI into the Xero platform, reducing manual data entry and saving businesses time.

The new feature uses machine learning to predict the contact and account code for transactions that cannot be matched to invoices or bills using an organization’s bank rules, Xero’s matching logic, or memorizations. Previously, businesses had to manually enter new contacts or account codes to reconcile these transactions — a time-consuming process that risks manual error.

Bank reconciliation predictions adds to a growing portfolio of Xero’s AI enabled product features, most recently Analytics Plus, a suite of planning and forecasting tools, powered by artificial intelligence and designed to help businesses and advisors plan for the future with confidence. Xero also uses two core machine learning techniques — text classification and entity recognition —to free small business owners and employees from repetitive tasks with automatic form filling in both Hubdoc and Xero Expenses.

Kendra Vant, EGM of Data said: “As we continue to leverage AI to build rich customer experiences, we’ll streamline more and more critical business tasks and processes for our customers, whether it’s reconciling transactions, filing expenses, or forecasting for the future.”

Harnessing AI forms part of Xero’s global data strategy, building on the data flowing through the small business platform to create new capabilities that improve the Xero experience for businesses and advisors, save them time and deliver insights to help them plan for the future.

Bank reconciliation is one of Xero’s most used features, with more than 1.7 billion transactions reconciled in the Xero platform over the past 12 months. The machine learning algorithms for bank reconciliation predictions learn from millions of these historical reconciliations across different organizations. As the algorithms improve over time, businesses can complete bank reconciliation faster, with more accurate information and reduced manual data entry.

“While each small business is unique, there are many patterns we can learn from the reconciliation activity of our millions of subscribers globally. This scale and reach allows us to tap into the ‘wisdom of the crowd’ and reduce toil for small business owners, while maintaining the security and assurance they expect from our platform,” Vant said.

“With these new bank reconciliation predictions, we can suggest to a user that money spent at an office supply store is likely to belong in Office Expenses, even if it is the first time you’ve shopped at that store.

“This may seem inconsequential, but any time spent on manual data entry is time not spent on the business. By streamlining a core task like reconciling bank statements, we’re able to reduce stress and give business owners more time in their day, while making sure their data is accurate and up-to-date so they can plan with confidence with their advisors,” Vant added.

The data team at Xero has expanded in the past two years and now operates from Australia, New Zealand and Canada. Xero’s team of data scientists and engineers work to integrate different AI applications across the Xero product suite, giving them opportunities to contribute to Xero’s growing portfolio of machine learning products operating at scale.

The new bank reconciliations predictions feature will roll out in phases to all customers. As Xero continues to improve these algorithms, users will see Account and Contact predictions in bank reconciliation more frequently.