Businesses are investing in RPA (Robotic process automation), but the results of some of their projects are far from encouraging: companies rarely implement more than five bots. What does the expensive and sophisticated equipment fail to deliver and why is it a good idea to invest in smart evaluation of business processes before plunging into large-scale automation?
From high hopes to disappointment
Not so long ago, UiPath – one of the largest RPA market participants – fired about 400 of its employees. Phil Fersht, founder and CEO of HFS Research, commented on this news:
“UiPath is realizing to its cost that intelligent automation is a marathon, not a sprint. It pushed the hype around RPA far too aggressively.”
The expert also noted that the RPA still has “some terrific benefits to support efficiencies and process transformations.” However, in his opinion, companies are now experiencing difficulties in adopting intelligent automation at scale – the vast majority have failed to surpass five bot implementations. Deloitte data confirms the estimate: at the end of last year, only 4% of companies used more than 50 robots.
At this scale, it is very hard for companies to see actual benefits of robotization. In most cases, they automate processes “on the surface” or the processes where using RPA robots would guarantee an advantage for the business. And when the project is complete, the legitimate question is: what’s next? When companies lack the understanding of their processes and lack a long-term automation strategy, it results in these five-bot projects. According to London School of Economics professor Leslie Willcocks, organizations get fast results but only see limited opportunities; therefore, RPA doesn’t go further than several pin-pointed applications.
Slowly but not always surely
“Spontaneous” robotization won’t work. For RPA projects to bring good results, it is extremely important that before buying software companies invest time and resources in evaluating their business processes. The more the organization knows about them, the easier it is to develop an automation strategy. Today there are three approaches to this in the market.
Some organizations invite external experts who can help to identify the processes where RPA will increase performance and reduce the impact of human factor. These consultants perform a task and then create heat maps that show efficiency of automation for a specific process. However, manual preparation of these reports is very time-consuming.
In order to make the work of the consultants easier, Process Mining solutions have been developed and employed. They collect logs – files that contain the information about users and programs’ activity. Then this data is sent to the team of analysts who review the information and then develop heat maps. But there is a problem: some of the logs are very difficult to figure out, and the Process Mining systems do not understand, for instance, what a user is doing in Microsoft Excel. Therefore, after heat maps are drafted and before business processes are developed, analysts have to review the accuracy of their reports with the employees, thus distracting them from their work. As a result, the Process Mining systems do not really simplify or speed up the automation. These solutions have good tools that help the company understand its processes, but they are not suited for the further robotization.
Process Discovery: All inclusive
The first two approaches: the manual analysis of processes and, say, semi-automated, do not provide the company with what it needs for the further successful automation: that is, the full understanding of how to evaluate their processes, the accuracy in dealing with information, and most importantly, operational efficiency. Vendors know the problem and are working on it: market participants have realized that the most promising way is to delegate process evaluation to IT solutions.
This resulted in launching a new class product, Process Discovery, which utilizes opportunities offered by the AI technology. Unlike other solutions, AI works here not as some sort of add-in for improvement but a proper part of the system. The Discovery bot is installed on the computer of an employee and runs without the latter’s noticing its work. Process Discovery bots scan only the employee’s workplace operations to gather information on the amount of time spent on a certain task, its frequency, and existing deviations. Then the system uses this data to promptly develop a working process and analyzes whether it can be utilized for robotization. After that, the AI gives recommendations as regards what processes it considers most prospective for delegating to RPA assistants.
The new approach is distinguished by its PD tools using on the same technologies as an RPA platform. This allows companies to spend massively reduced amount of time on switching from analyzing processes to their algorithmization and parameterization, and a user can start creating an algorithm for a robot upon immediately upon receiving recommendations from the PD system. This makes it much easier for a company to implement a large-scale robotization; a project does not get scrapped upon automation of several processes as the company has a clear idea of the remaining ones as well as the system recommendations, and is highly likely to advance.
According to Kryon experts, return on investment in such projects can reach 70% against 50% we are currently observing. This will lead to data mining becoming an important part of digital transformation in business. This is a hot topic, yet few realize where to begin – and now there is a simple solution to that.
By Andrei Kapranov, Director, Center of Innovative Technology, Navicon