Insight & Analysis

Why robots fail: best practices to ensure treasury automation strategies are successful

Published: Oct 2019

The robots are coming. But sometimes they just aren’t that good in a treasury context. We ask an expert why this is, and how to make improvements.

Two small robots, one with an error and the other with a tool trying to fix it

An increasing number of treasuries are turning to robotic process automation (RPA) to eliminate time-consuming, laborious, manual tasks. The common belief is that this can free-up staff to focus on other, value-adding, projects.

However, whilst many organisations are already reaping the efficiency and cost-saving benefits offered by RPA, some implementations have been disappointing in terms of results and ROI, claims Andrew Hayden, Senior Product Marketing Manager at US-based data management and process automation firm, Winshuttle.

More can mean less

The analyst group Forrester, predicts that the RPA market will reach US$1bn by the end of this year. It also expects that one-tenth of all new start-ups will begin life with more ‘digital workers’ than human ones. These days it is rare to find a large enterprise without some form of RPA initiative aimed at increasing business efficiency, so why are the outcomes sometimes underwhelming?

The answer is in the approach to automation, says Hayden. Despite RPA’s reported popularity, businesses often underestimate the complexities involved in putting robotic processes into practice, he says. “The organisations achieving the best results are those which take a holistic approach, thinking first about what processes they want to automate and then choosing the right tools for the job.”

It’s certainly a theory supported by analyst firm, Gartner, which advises that firms “do not try and make one RPA tool answer all your organisation’s automation needs”.

Getting specific

A ‘portfolio’ methodology, where specific RPA tools are matched to specific automation issues – in areas such as accounts payable (AP) and receivable (AR), bank account management (BAM), credit and collections, and FX – is particularly relevant when it comes to accelerating processes around complex systems such as SAP, Oracle and other ERP platforms.

These processes could include manual data entry, data extraction, or data movement during upgrades or transitions. “We see many instances where general-purpose RPA bots become the ‘square peg’ to SAP’s round holes, due to the proprietary processes and nuances of the platform that haven’t been taken into consideration,” comments Hayden.

Whilst these general-purpose RPA tools may be sold as ‘no-code’ solutions, most require a high degree of programming to fit specific situations, and they can be unreliable when applied to complex ERP processes. This, he believes, is especially true when automating SAP ERP processes.

Stepping beyond treasury, there are other significant challenges to consider when deploying RPA in an enterprise automation initiative. Some come as a result of RPA teams having to navigate the requirements of multiple departments and business units, while others are hindered by technology and compatibility issues. Governance and audit restrictions on how organisations manage data can also cause problems.

General errors

Additional challenges seen by Hayden’s customers when implementing their enterprise automation roadmap include the following:

  • Bots can fail or break: if a general-purpose bot was programmed via surface-level screen recording, simple changes in the environment can ‘break’ the bot. For example, frequent SAP GUI updates have been known to cause significant issues and breakages, requiring time-consuming bot updates or reprogramming.

  • Beware of replicating poor processes: it’s essential to ensure all processes are optimised before they are automated. Otherwise it will just mean that the bot is replicating a bad process rather than improving it.

  • Conflicting business metrics: whilst the RPA team is likely to be measured on the number of processes it automates, business teams are typically measured on results, data quality, or employee effectiveness. It’s essential that the two ‘sides’ collaborate to achieve both goals and for the project to be deemed successful.

  • Is automation really necessary? Not all business processes are ripe for automation, and sometimes RPA teams may just be magnifying the inefficiencies in the manual process, rather than improving them. Equally, automation might be a step too far, and a degree of pragmatism should always be exercised with regard to automation.

Right tool for the job

Taking a portfolio or holistic approach to RPA deployment in any treasury or finance function requires an examination of both the processes to be automated and the expected business outcomes, in order for their automation roadmaps to be developed, explains Hayden. “Treasury can then choose the best tool for the job – and a general-purpose bot might not be it.”

For example, SAP process automation solutions are built specifically for SAP processes. These allow non-technical staff members to record through the SAP GUI, while the actual execution of the bot takes place at the application layer.

SAP-specific RPA solutions also enable users to quickly intervene if data errors are detected, using familiar treasury tools such as Excel for data management. For Hayden, it is evident that applying the right tool for the job “results in faster implementation, lower deployment costs, and a higher ROI than experienced with generic RPA tools”.

The time and cost-saving advantages offered by RPA in a treasury context are plentiful. But for it to be effective, it’s essential for those using it to understand the strengths and limitations of the approach and the tools in their portfolio. As Hayden concludes, “when it comes to RPA, one size certainly does not fit all”.

All our content is free,
just register below

Already have an account? Sign in

Please only use letters.
Please only use letters.
Please only use letters.
Please complete this field.
Please select an answer.