A Three-Part Conversation With Professor Leslie Willcocks, co-author of ‘Becoming Strategic With Robotic Process Automation’
Do you see any international players that will be entering the league of these top three US players?
The top three RPA players are not US players, though they are very heavily into the US market. One is from Budapest – UiPath, Automation Anywhere is in India and Blue Prism has its roots in the UK. What’s more likely to happen, I think, is that they all get primed to sell at some stage. I don't think any of these three are currently at their full potential value. I think they’ve got another three years to fulfil their total potential value (sales value) because the market is still quite small and they’re growing exponentially. So why would they sell now? The premium price will come much later down the road, when they’ve done all their marketing and sales and they’re achieving high sales and customer numbers. By 2025, RPA alone could be worth over 15 billion US dollars revenue a year. The total RPA & cognitive market could be about 46.5 billion US dollars by 2025. I think one likely scenario is that one of the big US players – it might be a system integrator; it might be a Enterprise software company; it might be a management consultancy – would buy them, or at least one of them, depending on a fit with their own offerings and degree of complementarity in business terms.
Can you give some indication of market size?
All the figures are Disneyland to start with – including mine! You just have to accept that, but what we definitely know (from advisory work by consultancy companies and big players, etc.) is that, by the end of this year, the total revenues from the fifty-two RPA players will be about 1.5 billion US dollars. But we know that those revenues are increasing at about 48% per year, and I think it’s probably closer to 65% this year, and that will go on for at least three years. You can sort of multiply it up, so that by 2024, RPA will probably be about 12 to 15 billion dollars.
The size of the cognitive market is much more speculative, because a lot of it depends upon the ability of corporates to absorb these technologies. My view is that they can’t absorb RPA, let alone cognitive at the moment. They’re just sort of running out of capacity to absorb ever more technology. And again, it’s a very unsaid research fact that, when you look at over 400 deployments, you get the impression that they’re never going to do this in the time that they think they’re going to do it. They’re just not ready to do it at the speed with which people, analysts especially, foresee. The number of difficulties and challenges along the road that we have been documenting tells you why that might be the case. So 46.5 billion dollars revenue by 2025 for the RPA, cognitive and AI market is probably not a bad estimate – but it is still speculation. Most of the analysts don’t really know, but obviously they’re making quite a lot of money out of producing the reports and speculating about what the figures might be. As I said, I think a major drag on sales might be the absorptive incapacity of a lot of organisations. Not everyone proceeds at the pace of the top 20% of corporates. There are an awful lot of organisations that are just not proceeding very fast at all.
Can you clarify a little more the basis of these figures you mention?
There are basically four sources of revenues: There’s the RPA licenses revenue (which is quite small really): support services revenue; the advisory services revenue; but also a revenue from where an Enterprise software company, almost on a hidden basis, has automation buried in its offering. This, nevertheless, could be counted as revenue – where a systems integrator that’s using RPA within its package with a client, or a BPO service provider that’s providing automation but doesn’t necessarily tell the client.
I don’t have a breakdown, but typically a license would be. They all argue about “ours is cheaper than yours”, but they’re not always comparing like with like. A license for one bot might be anything between 8 and 15 thousand US dollars a year, depending on add-ons, so that would be the income. Blue Prism probably provides the most reliable assessment of revenues because it’s a public limited company. Its last statement, in 2018, was 55 million US dollars in revenues, under 1000 customers but no net profit. I would guess that, for 2019, that figure would be a hundred million US dollars, and a lot more customers. By mid 2019, Automation Anywhere and UiPath, who are not public limited companies, were saying they have 2,300 and 2,800 customers respectively. Now the question then is, how many licenses are they selling to each of their customers? The RPA vendors do not make such data easily available, but I suspect it best to assume lower rather than higher figures for 2019, as there is a lot of evidence that most RPA users have not been scaling as fast as many analysts, and the companies themselves, were expecting. So you can work out the sums if you wanted. I’ve never done that it because it adds up to a small amount of revenue whatever your assumptions.
So the revenues I mentioned are from RPA players who are trying to offer more than just the license for the RPA software. Now they’re providing additional services like integrated platforms, which customers are bound to buy because they want to do more with their RPA. Then there are systems integrators that orchestrate RPA, cognitive, analytics and AI. These don’t look like RPA revenues, but clearly if they’re using RPA and selling the services, some of their income has to be attributed to RPA. So Accenture, Wipro, TCS, GenPac, DCDXC, are all systems integrators that are generating revenues. Then there are the cognitive producers that sometimes building RPA capability to complement their cognitive technologies, like iPSoft and Arago. Finally there are the Enterprise software producers. These are the big players like SAP, SalesForce, Oracle, who are all developing RPA capabilities to include into their platforms. Presumably someone’s paying for that, so you have to include that as RPA revenue, to some extent.
How do clients choose between the players? Any new insights?
Well, there are many set criteria templates, and organizations are very good at customizing their own to fit with what they want for the supplier. This is unpublished research by us, and to some extent speculative, but we got interested to find out who bought the different technologies. First, the technologies are not necessarily the same. Everything that’s called ‘RPA’ is not the same as everything else, and there are really big differences between the top three, although their marketing suggests they’re all doing the same thing. So the interesting question that we asked ourselves in this particular piece of work, was who, psychologically, was attracted to Uipath, Automation Anywhere or Blue Prism – we have only looked at those three at this stage.
What was interesting was that we identified the risk averse companies: these wanted reliable platform; that wanted audit trails; they wanted regulatory compliance; they didn’t want business units to declare UDI from the IT function, but saw it as a business imperative to automate in order to become more effective. These companies tended to favour Blue Prism because they sold Enterprise software. It was a sort of total package, solved most of the problems quite early on if you were willing to spend the time at the front end.
The companies who liked UiPath tended to be more technologists, engineers, who liked to tinker with systems; do a bit of coding themselves – liked playing with the technology; liked the open source code that UiPath used; recognised that it wasn’t the total answer, but wanted to contribute to the answer and customise it a lot more than you would be able to do with Blue Prism.
Those that went with Automation Anywhere tended to be people who wanted to declare freedom from the IT department, not have to get into the IT queue where they would be very low on priority because their need was very small and coding a low priority piece of software was not really the IT department’s game. So companies like Automation Anywhere came along with desktop robotic automation, assisted automation and provided something that you could be utilising quite quickly on a desktop. So people who wanted that sort of capability have tended to buy Automation Anywhere – it’s cheap, it’s quick, it gets you doing stuff, it gives you some levels of productivity.
Now all three have moved to the same ground, saying that they all do what all the others do, plus they have an automation platform that integrates with cognitive technologies. That’s an interesting claim, which will be proven or disproven as people buy it and work through it. I’m not convinced that they are all up to speed on that yet, but I’m sure it will come together in the next 18 months.