Xero recently introduced the accounting profession to the concept of no-code accounting, and expanded its mission to include the automation of bookkeeping.
“As a company we are focused on eliminating coding,” CEO Rod Drury told 2,300 accountants and bookkeepers at Xerocon South in Brisbane, Australia. “For millions of small businesses, we can just get rid of (reconciliation). We are signalling to the accounting industry that what’s just over the horizon is already here.”
Artificial intelligence and its potential impact on white collar workers has been a favourite topic of futurists for some time. Yet this was is the first time it has appeared as a feature in accounting software.
This article explains how machine learning (the technology behind no-code accounting) works and the impact it will have on accounting software. It also contains suggestions on how accountants and bookkeepers can prepare themselves for no-code accounting.
What is machine learning?
Machine learning is a form of artificial intelligence and is defined as "the ability for a computer to learn without being explicitly programmed". A computer usually does this by scanning large amounts of data, finding patterns of behaviour, and then creating an algorithm to copy those patterns or use them to predict outcomes.
We already use machine learning in our daily lives. Drury gave his experience with Uber. The app predicts to the minute the arrival times of its drivers by measuring traffic flows in real time.
Xero and other cloud software vendors already have huge amounts of data to train Amazon’s algorithms. In the past 12 months Xero processed over a trillion transactions and has many terabytes of financial data sitting in Amazon’s data centres.
This is why Drury spent so much time talking about the quality of the database behind Xero’s software. Drury says Xero has a deeper understanding of financial data due to features built into the platform such as double-entry accounting, multi-currency and standardised report codes.
Machine learning can use algorithms to predict someone’s intent; in other words, faced with this transaction, what is the most statistically likely course of action? The algorithm can then tell Xero to code a transaction in a certain way.
We're not just looking at what one person did, we're looking at how hundreds of thousands of small businesses work. Machine learning is how you build the magic. –– Xero CEO Rod Drury
The brilliance of machine learning is that it never stops learning; if the algorithm gets it wrong or doesn’t know how to code the transaction, it will record how the user coded it manually and then adjust its own set of rules.
The wisdom of the crowd
Xero is taking two approaches to machine learning. The first, led by Amazon, looks for patterns across large sets of data. The macro-trends, in other words, that create general rules about how businesses do their accounting.
Amazon’s machine learning operates best at that macro level, but not as easily inside individual file. Xero has also created its own algorithms to work within clients’ files and find patterns that each client follows. Xero can then create specific rules just for that client.
Xero has about 10,000 customers per database, and three databases per server. It is using Amazon’s technology to look for patterns in that data. Drury calls this “the wisdom of the crowd”.
A good example of this is the data from Find and Recode, a tool that accountants and bookkeepers use to re-assign large batches of incorrectly coded transactions.
The results showed that business owners make many basic mistakes which a machine would not make.
For example, the most common error was coding expenses as Sales in the general ledger – just because Sales is the first entry in the dropdown menu for account codes in Xero.
The second most common correction was re-coding transactions business owners had labelled as Office Expenses. Third was bank fees and fourth owner drawings.
Drury cited these results, drawn from three million transactions, as evidence that small business owners find accounting really hard.
Bank reconciliation is an obvious area of focus. Xero has a set of transactions across all its customers and can see the same suppliers (such as utilities) and track the different merchant numbers for every bank.
Another area of focus is the chart of accounts. Xero can spot consistent patterns in coding transactions from certain suppliers. A transaction from FaceBook will almost always be coded as Advertising, for example.
Machine learning can already predict this and Xero wants to automatically code those transactions on behalf of the business owner. The company processes 44 million lines of bank transactions a month and when it ran machine learning over the data the algorithm correctly coded nine out of every 10 transactions on its first pass. “Bank rec goes away,” Drury says.
In fact, Xero is so confident that machine learning will code at a higher rate of accuracy than business owners that it wants to turn the chart of accounts invisible to them.
In the next version of invoicing in Xero business owners will no longer have the ability to code the invoice, Drury says.
How can you prepare for no-code accounting?
The best way to tackle technological disruption on this scale is to disrupt yourself first. Swallow the bitter pill before circumstances force you.
If your revenue is heavily weighted towards bookkeeping and compliance, automate it as quickly as possible. The first steps to automating include mapping out your processes and working out which ones you can replace with cloud programs.
You should also become experts at using bank rules. Most people only set them up to do basic activities but they can be really powerful if you know how to use them.
(If you need help in this area, check out our guide Top Tips for Bank Rules, drawn from the combined wisdom of surveyed Digital First readers and the Australian Bookkeepers Network.)
Automating compliance should free you up with more time and give you better quality data faster. This creates a strong platform to offer other services such as management reporting or virtual CFO, wealth management, and so on.
No-code accounting will soon be a reality. Technology marches ever onward.
Move yourself into position where this becomes an opportunity for your business, rather than a death sentence.