Accountants will have to learn to follow the logic of the algorithm

It won’t come as a surprise that knowledge-intensive sectors are the ones with the highest potential for digitalisation. This was confirmed and strengthened by the recent study from McKinsey and HBR which showed that IT, media, professional services and finance & insurance were the four industries with the highest digitalisation rate across a wide range of industries and features analysed.

Accounting and bookkeeping – part of the professional services industry – is definitely not an outlier here. It is all about processing information (from a source document or transaction to the ultimate report, be it management, financial or tax); therefore it is all knowledge based. And nearly all of the necessary information comes already digitised or can easily be made such (by scanning and OCR) with a noticeable exception of certain non-standard contracts which may still require some interpretation and extra manual work before they are actually recorded in the ledgers (such as M&A contracts) to the extent they cannot be standardised.

Given all that the technology development had to have an impact on the way accounting and bookkeeping is done. And – it will have even more within the coming years when specific accounting-oriented applications of those developments emerge.

From what we already know, the following technologies have been revolutionising the industry over the last years and will continue to do so in the future:

Robotic Process Automation (RPA)

The original thinking that a robot must be a human-like creature performing tasks with metal arms has been overthrown with the concept of a software robot – a piece of software that performs task using a physical or virtual computer. This piece of software is in fact an equivalent of a person (in this case say – a bookkeeper), performing repetitive tasks by logging into applications, retrieving data, transferring it to other applications, generating reports, searching the web for necessary data (such as. downloading bank statements).

Incidentally – the use of RPA removes the need for costly software integration which could do some of those things as well. Here – we need no integration. All we need is a user account in a variety of apps, and that user moves around those apps juggling inputs and outputs. And – the user would never get tired, and can work night shifts without complaints.

So for manual repetitive tasks with large volumes (yes – those are necessary, because the robot works faster and has no coffee breaks) RPA offers a sound solution at an affordable price (equivalent to more or less one FTE – but available 24/7).

The downside of this technology is that any process delegated to the robot has to be very well documented and stable. When a process changes – the robot has to be re-programmed – unfortunately in this case, briefing it on the changes to a procedure in a social room during lunch break is not an option.

Machine learning and tacit knowledge

Although many of my fellow accountants would try to disagree, in my view 80% of transactions are fairly standard or can be standardised (after all – even if you are a venture capital firm, 80% of what you do is investing and divesting which then becomes fairly standard and even if individual contracts differ – they all contain the same key points – the problem becomes then to extract those points from the contracts).

If so – even if we do not have very well established procedures for each standard transaction (we have the so-called tacit knowledge: the knowledge is held by individuals in the organisation not by the organisation itself), we may employ machine-learning tool, which would determine the accounting classification of various transactions by tracing them back from the outcome (information segregated on ledger accounts) to the original document. What we need for this is: the tool (available in many forms from many suppliers), and a large chunk of data (transaction data from prior periods would be excellent and would provide necessary volume to teach the tool). And of course – a specialist who could link one with the other.

Based on this, we may have an automated system for posting all repetitive transactions (not already dealt with automatically like bank statements).

Both of these technology developments are already here and now. It is not a sci-fi prophecy. Given the technology in our hands, we are able to eliminate most of the tedious, repetitive, or burdensome accounting tasks either through RPA or machine-learning algorithms.

Due to the specifics of their nature – it is easier to make use of them in large organisations with millions of data items and stable procedures. However – make no mistake – these solutions will find their way into a middle market and ultimately – even to sole proprietors. How?

Well – for the middle market – small and medium-sized companies – the outsourcing service providers will have the necessary volume to use the benefits of automated solutions, by combining data from various sources and various clients. As a result – the outsourced service will become more efficient and more error-free (as human errors would be eliminated). When the potential for efficiencies is noted by the regulators, the regulatory reporting schedules would be tightened for all entities and management would also require information sooner to keep up with the competition. And then – most likely – the outsourcing providers would replace in-house accounting departments when those would not be able to meet deadlines.

And finally for micro-entities: again the solution will be based on combining the data and processes of a large number of businesses (possibly segregated into silos for various types of activities) and automation will be generic to each silo. It’s easy to imagine a fully automated process for tax reporting for those entities which usually do not even do a double-entry bookkeeping.

Overall – there will be no more unskilled or low-skilled bookkeeping jobs in finance departments. Instead – model designers, data cleansers and ML model testers will be in demand.

Another game changer that will impact the accounting industry is the big data coupled with increased computing power and machine learning (again). In this context big data would also mean non-financial information that would and should be captured alongside numbers. The modern cloud based, FinTech-developed reporting tools accommodate information from various sources and allow its presentation in formats and breakdowns which add new dimensions to decision making.

Currently the reports are designed by finance people. But once these tools are reinforced by machine learning that determines relations between data items which were missed by humans – the management reporting would become of even greater value to decision makers.

Thus – management accounting departments will gradually move towards designing new reports as per regulatory and management needs and creating information from a much wider ecosystem of data available to be processed and analysed.

Distributed Ledger Technology

Finally – the Distributed Ledger Technology (one example of which is blockchain) is another tech trend which is said to have a big potential to impact our accounting departments.

The big advantages of DLT include its security (copies of all historical transactions are stored in each node of the distributed ledger and any change to it must be negotiated between the nodes) and transparency (anyone with an access to the DLT can view all transactions and changes).

Currently accounting systems are built centrally – there is one general ledger and even if the actual transactions are posted all around the world – the change happens in one central place only. This has advantages, but this solution’s one big disadvantage is its susceptibility to a potential fraud – if books are to be ‘cooked’ – you need formal permission in one place only. No independent node would have any say in the process. Similarly a hackers’ attack on company’s accounting records could be targeted at one place only. Yet a DLT requires approval from a wider group of nodes to effect the change; attacking just one node would be unsuccessful.

From this perspective, maintaining books of accounts (actually ‘ledgers’) in the DLT makes a lot of sense and therefore it may be expected that gradually the centralised databases of ledgers would be replaced by DLT solutions. My guess is – listed companies would be the ones to pave the road for it.

Increased transparency will be another argument here. Assuming their access to the ledgers, the auditors or regulators could verify each transaction in any node of the DLT being sure that the entries for the transactions had not been modified since their recording.

New developments in technology are having or are about to have a major impact on the shape of accounting departments. Bookkeeping, accounting and reporting will become more automated, triggering changes in staffing. There will be less space for traditional bookkeepers. Transactions will be posted, classified and reported automatically.

Large volumes of data will be added to the current financial information for decision-making purposes and reports will be generated – and sometimes even designed – by machine learning. And machine learning (or what is now too often referred to as artificial intelligence) will analyse and interpret the data, since it proves far better with handling such large volumes.

This again will have an impact on the people we will be looking for in finance departments – those will need skilled visionaries capable of inventing new data sources to support reporting and diligent analysts that would attempt to follow the logic of the algorithms to detect potential errors in reasoning or simply provide the audit trail.

All of those changes will take time. They won’t happen overnight, but today’s accountants must be wary of the fact that their profession evolves, probably at a faster pace than many others – simply because it is all about information processing. And they need to adjust and upgrade.