Interview of Professor Leslie Willcocks by Professor Christian Langmann of Munich University of Applied Sciences
The interview was published in: Rethinking Finance Magazine (www.rethinking-finance.com), June 2020
CL: You were one of the first academics publishing research on RPA. When was the first time you heard about RPA?
LW: I heard about it in 2012 when the notion of robotic process automation was promoted and claims were made that it could achieve very large benefits quite quickly. But I put off studying it for a long time because every year my co-workers and I get together and say, well, what are we researching this year and what new themes are there that where we can investigate? We wait until we find themes that are really interesting and managerially effective with business benefits. There has to be something there.
So, we really had to wait until 2014 when a vendor approached us and said, well, we think we have those case studies. Would you like to look at them? And I said, I don't want to talk to you about them I want to talk to the clients about them. They gave us access to the clients. That resulted in finding out what RPA exactly was and producing our first book in 2016.
There was something there, but we recognized that it was a very small market. There were very few small vendors in the marketplace and very little take-up by corporates and not at all by government agencies.
CL: What was the vendor’s name? Was that Blue Prism?
LW: Yes, they invented the category because they had the software since about 2003. I think, they applied it originally to a German utility. They learned an awful lot because the German utility is very strong on “Does it match regulation?”, “Does it match with our enterprise infrastructure?”, “Is the risk low?”, etc. That was their origin in RPA, but they never called it RPA until 2012. They had quite a number of quite large clients by then. But it had no profile whatsoever.
CL: When you talked to the first company, what were your first thoughts?
LW: I did not understand it and thought “you're not marketing this very well”. This is quite understandable because it's difficult to explain RPA initially, to try to discriminate it from how other IT-systems work and persuade the IT-function that it is different.
The vendors had not really gotten their marketing act together by then. They had not put it down into words that were easy to understand by corporates or business users or IT. It all seemed a bit like a fairy tale: How could you do it so quickly, so cheaply outside of the IT function, and get some results? There was a sort of low-profile disbelief factor with it for about a couple of years. Then the marketing was confusing because they all claimed to be doing the same thing. And in fact, RPA comes in many different ‘flavours’. About over 55 vendors are marketing something called RPA, but most of those products are not the same thing.
CL: In your view, how has the market for RPA changed since then when you first elaborated on this topic until today?
LW: A number of things have changed. First of all, the profile has been much higher because certainly the top three vendors have invested a large mass of money in marketing, sales, selling staff. Secondly, the product has moved on in the sense that originally it was sold as a fairly quick-win, tactical technology tool that would boost productivity and would solve some of the problems that were ongoing without too much hassle, without a complete overhaul of the legacy systems, the infrastructure and all the rest of it. It was suddenly recognised that you could apply it to any sector and any process as long as the process had certain characteristics. The three leading companies have also now invested in the link between robotic process automation and the next stage of cognitive automation tools. That issue is what I call the ‘next phase’, where they connect RPA with augmenting cognitive automation, usually provided by different vendors.
I think competition has helped amongst the RPA vendors. Also, venture capitalists have held on to this as a breed. They say, “we’ve got a lot of money sloshing around and we have a lot of technologies to invest in. AI looks like a really good one and this may be sort of low down on the food chain of AI. But it looks like AI. It sounds like AI.” And the market penetration is so low that it has massive potential. You have got a market now. I think about US$5.5 billion revenue in 2019, for RPA cognitive and AI suppliers combined. This is not huge at the moment but when you see what RPA can connect to, the potential is massive.
So, venture capitalists look at that. I think, also corporates have looked at this: “Well, we really do have problems, can we marry it to our existing information systems or if case studies suggest we can, we ought to be getting into what is called AI as an umbrella term for this information technology. We need to make a start and RPA is the obvious place to start.” So that's a pretty good summary of from where we came and how we've got here.
CL: Currently, we continuously read about the introduction of RPA in many global corporates. However, small and medium enterprises (SMEs) are often still unfamiliar with the RPA technology. What is your view on this? Why do you think SMEs still struggle with RPA?
LW: The big game where these automations pay off is where there is scale, complexity and the need to reduce the time to market or time to do a process.
Small companies don't have scale. Generally speaking, they do not have complexity and they don't have a bunch of money to spend on any other technology, but they have some sort of rudimentary ways of doing the process anyway. And they probably do have problems, but probably they can solve them more by utilising human labour and existing technologies than getting into RPA. I also think, that corporates have a problem because the ability to innovate, to bring in new technologies is declining slowly because they've got so much else to do, so many other priorities that when you say we are now going to do automation they suddenly don’t aim further budget on it and it does not get senior executives’ attention and the change management component is under-resourced. They do not understand that to do this strategically, in a big way, you have to really throw a lot of resources at it. And so, it gets under-resourced and as a result underperforms even in the big corporates, which is why, you have figures like only 13 percent that can actually scale to any great degree in terms of robots, and recognising that it is an enterprise resource, not just a desktop one.
CL: RPA is ideally suited to fix fragmented IT-landscapes with poor interfaces. Therefore, some companies believe that the introduction of RPA is leading to a slower transformation of legacy systems. How do you see the relationship of RPA and the use of legacy systems?
LW: This is an interesting point. The view is that it is a quick solution to some intractable problems of the legacy and that that will delay developing the legacy systems, because it only partially solves the problem. I think a number of things are going on there. First of all, all these corporates are supposedly moving to digital transformation and emerging digital technologies. They should be developing their legacy systems anyway to take advantage not only of RPA and cognitive, but also social media, mobile cloud, analytics, blockchain, Internet of Things and augmented reality. All these Internet things should be on the corporate agenda. I'm not quite sure why they are looking so far backwards at their legacy systems because they should be changing it anyway. To me, this technology is part of the change, the digital transformation.
CL: When talking about implementing RPA, it seems as if there are two views. One view discusses RPA very technically. But there is also an operational view of implementation. What's your experience with the implementation of RPA?
LW: You're very accurate. Most of the failures are not failures, but disappointments, troubles, because they haven't put in place what many people call a robotic operating model. I would call it governance, change management—all the things that we cover in our book ‘Becoming Strategic With Robotic Process Automation’. We cover that and the reason why we do is because we found in our research looking at hundreds of deployments, that 75 percent of the problems are managerial and organisational and only 25 percent are technical. You will get technical challenges and some of them are different depending on the vendor and some of them are more costly to implement to enterprise level than others.
But if you have the governance in place, you have a very strong basis for implementing RPA. Advisory companies have their governance models, as well as some vendors. I certainly would make sure the governance straitjacket is in place very early. If you do not do that, if you take the hype from the salesmen and say, “oh, we just bring it now from the cloud and off we go it works off the mobile as well.” Unfortunately this is true only at the most rudimentary level. It's an awful lot of work behind this before you can actually utilise it beyond desktop automation to enterprise level.
CL: Let’s assume you accomplished the RPA implementation successfully. In theory, RPA frees up capacity of employees for other, more valuable tasks. But how does this look like in reality? A survey recently published, for example, could not really prove that theory.
LW: That is very interesting. We have been looking at successes and failures. I'll give you some figures there because they are interesting to structure the conversation. We found, for example, that really only 20 percent of organisations are particularly successful with the RPA technology. Then we've got 25 percent that were trying to get there and doing so many things right but some things wrong and ran into challenges because they have still got a lot of work to do.
But then you've got 55 percent, and the bottom 20 percent of those 55 percent are failing. They don’t do it at all. And then the other 35 percent are really struggling, especially on the management front.
That helps to explain the sort of survey results you get. Because in that structure, you would expect only a minority of organisations to be getting employee value out of this in the terms of employee satisfaction, because only they would have implemented it properly to get multiple business benefits.
The other thing is that we've got multiple case studies where we haven’t got these conflicts. We have interviewed the employees and they have said all the things that we report on this. We have been to lots of organisations where they said, “where it helps is that we have had great increases in the amount of work to be done, and it really helps us with this.” Indeed, they don't say to cut headcount. “The dramatic increase is in the amount of work to be done. We can't resource it. We haven’t got the skilled people. We have a very high labour turnover in certain jobs. They are very unattractive. And this helps us. And we have to do it because it's easy enough, it is more accurate. We have a number of benefits.” Employees say, “that's great because it takes those really awful jobs off me so I can do the other work, which isn’t necessarily great work either.” But the stuff that was really pulling them down was the repetitive ‘robotic’ work as opposed to the productive work. There is also a species of work that is highly distracting, needs to be done but not particularly productive. I call it the audit, regulation and bureaucracy work. It has increased massively in the last decade or so. RPA etc helps with that too—for example think for a moment about regulatory compliance in banking and insurance.
CL: What would you recommend to future students or employees? Do you think it's worth looking at RPA?
LW: Oh, absolutely. I have got some present students with masters degrees who have joined vendor companies because it's an exponential growth rate sector for at least the next three to five years. And it connects up very quickly now with cognitive developments. They are going to be in AI. Corporates eventually are going to adopt. And it gives students a really good knowledge base and experience base for the future. So, it's a good way in. And of course, the leading vendor companies are desperate to recruit.
So, I think it's a very good area. I think even if you're not going to be technically interested, it's very useful to have a technical understanding; a technical set of skills to put beside your interpersonal and business skills in the future, because these emerging technologies are going to be very important for the organisational functioning from now on.
CL: Consultancies or/and vendors think that the RPA market will explode in the coming years. They expect a combination of robotics with AI in the next phase. What is your opinion on this? How do you see artificial intelligence play into RPA in the coming years?
LW: It is a two-track game really. One game is the technology and the other game is how fast can organisations adopt and adapt to it. For the latter, there is a ‘fast track’ and there is a ‘it is not as fast as they say it is track’ and a ‘slower than they would like to present themselves track’. The ‘not so fast as they say it is track’ is that there are some impressive pilot uses of technology, modern automation and some impressive pieces of software and it doesn't add up to a joined up integrated automation capability that can be applied to most corporates around the globe. Where we are is a very small market. By 2024, a prediction I've seen for the size of the robotic cognitive and AI market is only US$46.5 billion revenue. It will be interesting to see whether the pandemic we are witnessing will accelerate or slow automation growth.
The IT and business process outsourcing market was US$1.6 trillion in 2019. Comparatively, RPA, cognitive and AI is a very small market, despite its huge potential. These figures do not suggest that the take-up is going to be that fast, or even impactful. And I think one of the reasons is that companies are even struggling with just RPA. So why are corporates going to suddenly discover how great it is, let alone start implementing even more sophisticated, cognitive technology and AI? In my head, knowing what I do about the technology and the area of IT development, and remembering that to develop technology that works in the lab is one thing whereas commercially is quite another… in my head I think there are going to be lot of challenges.
I think, the fast track—the speed at which technology will develop—is not so fast as it says. The organisation track is a lot slower. I've just written an article called ‘Slow Train Coming’, which is based on investment banking, and the technology I have been studying is beautiful. But the ability to absorb it is not beautiful at all. What was pulling them back is what I call the ‘seven siloed organisation’. You mentioned earlier that there were legacy technologies and there's also legacy data and the way that it has been poorly managed. That's a great problem when you're trying to get into cognitive technology, especially as RPA can only use structured data and 80 percent of the basic data in an average organisation is unstructured. This is a big drag on speed.
If you've got data and technologies, that’s two silos. You then have organisational silos in skills; culture; process; managerial mind-sets; and structure. I identified that there are seven serious ones when you go into organisations. Whenever you ask organisations how siloed they are on a range of 1 to 10 the average comes out at about 6, and not just one or two silos, but across all seven silos.
Summarising this, the fast track is not fast as fast as everyone says. And the slow track—the organisational track—is a slow train coming. McKinsey produced a nice figure looking at how technology is deployed and found that it takes between eight to 26 years until 90 percent of businesses deploy a particular technology across the globe, and I think that's pretty accurate. I've seen it happen many times. There is no reason why automation should be that different. Having said that, many already believe that the pandemic will provide a burning platform for much faster wider deployment of digital and automation technologies. I think organisations will react to where the business pain points are, and many will be trying to get back to financial basics, but much depends on whether they take a short-term crisis view, or a more strategic view. Automation will be adopted as a quick cost saving fix in desperate times, and we will also see, elsewhere, an acceleration in the existing trends towards digital technologies and developing digital businesses. But those 20 percent digital leaders we have identified are going to ride out of this much better than the other 80 percent; I think that’s a fairly safe prediction.
CL: Let’s finish with a question on the future. Do you think that RPA will still be there in 25 years?
LW: RPA will not be talked about in 25 years. It might disappear in 5 years actually. These are speculations, not predictions. If you look at major vendors, what are they trying to do exactly? I think, ultimately, they are trying to create as much business value as possible in terms of their own value, with the goal of eventually being sold, because they are very small companies. And so, you can see that at some stage, the big players, such as Microsoft, Amazon, Google, Oracle or SAP, are going to look at this and say, we need this because as it is a necessary technology, but do we need to start it as a high-profile thing or do we have more interesting things that we should be trying to market. Over time it will be necessary, but invisible.
CL: Thank you very much, Leslie, for giving us insights and your view on RPA and its development in the future.
Leslie Willcocks is Professor of Technology Work and Globalization at the London School of Economics and Political Science.
His latest book ‘Becoming Strategic With Robotic Process Automation’, is available to purchase onlione at www.sbpublishing.org.
Leslie’s forthcoming title 'Global Business: Strategy in Context' will be published later this year.