RPA can be implemented by any organisation with a digital infrastructure, from SMEs to conglomerates.

According to research by RPA provider Softomotive, 46% of medium-sized businesses in the US have employed RPA at scale across multiple parts of their organisation.

And it’s not just the private sector that is benefiting from automation software.

Governments, especially those in times of austerity, are using software bots to pick up the extra workload created by reduced budgets.

In March, for example, the UK’s National Health Service announced it would trial Automation Anywhere’s RPA at one of its trust’s corporate services arm.

Two years ago, it was predominantly the banking and finance sector embracing RPA, says Dening. But the market has opened up to cover almost every industry: manufacturing, healthcare, mining, insurance – the list is endless.

Blue Water Shipping, for example, a Global logistics firm, originally stress tested its online order system manually, before switching to automating it using RPA vendor LEAPWORK.

“Manual testing was just too repetitive and so boring,” said Ronni Sohl, a Blue Water Shipping test manager.

Manual testing – including variables such as item number, cargo type and transportation mode – took a full day, but LEAPWORK’s RPA reduced it to 30 minutes, the company says.

Determining the price of an RPA solution depends on the number of bots and software components that make up their RPA deployment. But, on average, a single bot (or unit) tends to cost somewhere between $5,000 and $15,000.

The UK government has a subscription with digital transformation firm T-Impact to provide RPA as a service and automate public sector workflows. As part of this, it provides software bots from US RPA vendor UiPath. These are priced at £12,750 per unit. According to T-Impact, it reduces UK staff costs by 90%.

Price will largely depend on the complexity of the task for the software bot to perform. But once that bot is paid for, it can perform infinite tasks without extra cost.

“Pricing is agnostic of industry or volume of tasks performed,” say Softomotive.

Although the Softomotive customers can make use of that particular version of the software forever (a software bot’s not just for Christmas), they will have to pay an annual fee to receive upgrades.

The savings from RPA will, of course, vary from business to business. The simplest rule of thumb is to calculate the cost it would take for a human to perform the same number of tasks.

According to Deloitte, a company deploying 500 RPA bots would cost around $20m. Assuming the bots replaced 1,000 employees, that could create as much as $100m in savings.

There are also ROIs to consider that are more difficult to put a precise figure on. For example, human staff with more free time could develop creative projects that generate more revenue.

This extra scope for growth, alongside improved employee satisfaction and better customer relationships, may be harder to quantify but should all be taken into account when deciding if RPA is a cost-effective investment for your business.

Given all of the pros to RPA, you might wonder why all companies haven’t implemented it yet. That’s because there are still barriers, such as the initial cost. While this is likely to be small change for larger companies, smaller companies might struggle more to get the expenditure signed off. And in some cases, there just won’t be sufficient ROI to justify the project.

Added to that, survey by EY found that 30 to 50% of initial RPA projects fail, which may cause some to pause.

One of the simplest barriers is a lack of awareness and education around RPA. Part of that is down Automating workflows is by no means new, but it is only in recent years that the technology underpinning it has become more intelligent.

Some companies, particularly their IT departments, can be resistant to change, too.

Understanding its limitations, will mean less projects will fail. For example, good AI is dependent on good data. But across a business, documents might not be in an accessible, easy to read format. What might be clear on a PDF for a human could be more problematic for a software bot.

However, that’s changing, says Dening, with the advancement of cognitive AI. It’s an approach that uses tools such as document recognition and natural language processing and with software that “learns as it goes”.

If, for example, it struggles to recognise a PDF expense report that is formatted differently to the usual documents, it can be trained to recognise it the next time around via machine learning, building its knowledge bank up over time.

The key is understanding that RPA is not a silver bullet, it is merely a cog in the wheel of your business that, if oiled properly, could turn far more efficiently.