After just one invoice, machine learning techniques understood invoicing behaviour in general better than Xero experts. By the fourth invoice, early machine learning implementations are accurate over 80 per cent of the time and by the 50th they consistently reach over 90 per cent
Xero says new automation system is the first example of personalised machine learning in a small business cloud accounting system.
Developed in-house by a specialist team of engineers working full-time for a year across three different locations, the system uses detailed statistical analysis to learn from and assist the individual business and their partner based on their own specific circumstances.
The automation will mean small businesses no longer need to worry about where their invoice is filed - an invoice for time spent on site should be recorded against “Sales - Labour”, not “Sales - Materials”, for instance.
The machine learning automation evolves with the processes used by the business and their advisor - when the small business comes to create their next invoice, Xero automatically suggests the account code so they don’t inadvertently make a mistake.
It will initially be made available to a group of small business customers and their accounting partners for testing, before being launched to all Xero customers later this year.
“Simply talking about machine learning isn’t enough. The truth is, it’s really difficult to do beautifully. While many claim to use machine learning principles, actually implementing them across complex platforms to address customer’s real needs takes intricate, specialized knowledge of the technology, the platform and the user,” says Luke Gumbley, Xero business product lead.
“Our machine learning automation will give each Xero business and their advisor the unique insights and processes they need to cut the burden on manual tasks and let them spend more time growing their business,” says Gumbley, in a statement.
We’re not asking them to learn something new - we’re saving them time and money by ensuring the system learns from and for them
“We’re not asking them to learn something new - we’re saving them time and money by ensuring the system learns from and for them.”
Xero says it set out to tackle a common problem among their customer base: account codes. There are more than 10.1 million unique account codes in Xero created by small businesses, meaning items are often entered incorrectly, creating hours of work for the accountants who need to correct them.
Small businesses take an average of 1 minute and 38 seconds to create an invoice in Xero — that works out to 13,600 hours across the 500,000 invoices raised in Xero each day. For every second machine learning helps to shave off that average edit time, Xero’s machine learning automation will collectively save small businesses around a working month every day.
Learning likely to lead to further transformation
Although the true impact of the technology will only become apparent when the product goes live, the benefits have already been demonstrated in early testing, says Xero.
After just one invoice, machine learning techniques understood invoicing behaviour in general better than Xero experts. By the fourth invoice, early machine learning implementations are accurate over 80 per cent of the time and by the 50th they consistently reach over 90 per cent.
“We expect the wider adoption of machine learning to lead to further insights and improvements that will help small businesses, their advisors and other partners fundamentally change the way they work.
“In the future, Xero will be able to use aggregated, non-identifying data to identify trends, patterns, behaviors and industry insights. This is just the beginning,” says Gumbley.
Send news tips and comments to firstname.lastname@example.org
Follow Divina Paredes on Twitter: @divinap
Follow CIO New Zealand on Twitter:@cio_nz
Join the CIO New Zealand group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.