Process mining isn’t necessary to bot automation — but it can create a more efficient bot strategy.

Process mining is the capture of insights in order to discover, monitor and improve real processes, rather than assumed processes, “by extracting knowledge from event logs readily available in today’s information systems,” according to global technology research and advisory firm Gartner. Process mining tools automate the identification of processes, also called “process discovery,” recording them and checking conformance once the automation is in place.

“Process mining is a fundamental part of creating visibility and understanding before you automate,” Gartner advised in its September 2020 report, “Market guide for process mining.” That’s because process mining identifies the process before its automated, allowing banks to be sure of what they’re automating before they spend money to potentially automate a bad process.

Here’s how banks can use process mining to build a better bot strategy.

1. Identify processes ripe for automation

Banks can use process mining to identify and understand the thousands of processes within their organizations, said Alex Day, senior vice president at process management company Signavio.

Often, banks and other organizations make wrong assumptions about how a process works, and this can lead to automating a bad process, rather than the desired process.

“If you are automating bad processes, or processes that are broken, all you’re doing is making a broken process work a little faster,” Day said. “With mining, you can go in and make sure that the two things are aligned perfectly, and that all the processes you have in place are the right ones, and that they’re working effectively for not only internally at the bank, but also for the end customer.”

2. Prioritize bots by business value

With process mining and discovery of processes, banks can identify average process completion time, run time, and other key performance indicators (KPIs) for the process. This allows banks to prioritize bots based on which processes will yield the greatest impact, said Ziv Ilan, the professional services team leader for process discovery vendor Kryon, during a recent webinar on robotic process automation best practices.

It’s not unusual to incorrectly estimate how long a process takes to complete, which can lead to poor automation choices, Ilan said. For example, a manager may estimate that a process runs 2,000 times per month, when in reality it only runs 1,000 times. Not only does the bank not achieve the promised return on investment (ROI), but wastes valuable developer time that could have gone to a higher value project.

A company that recorded processes in parallel across various departments could decide where to automate based on data and business goals rather than what employees thought might be the highest value process, therefore arriving at “a new process that is better aligned with your KPIs, better aligned with your strategy in the company to streamline or improve the employee onboarding process,” Ilan said.

3. Identify sub-processes for a clearer ROI

Calculating ROI for a bot isn’t always straightforward. This is especially true when regulation and compliance are concerns because humans are more likely to be part of those processes.

However, process discovery can identify sub-processes that can be partially automated, and also help banks better identify how much money the bots are saving. ROI for unattended bots is calculated differently than for fully attended or partially attended bots, Ilan said.

4. Optimize infrastructure

Mapping out processes also allows an IT team to identify exactly when processes aren’t running. Perhaps a process isn’t running during employees’ lunchtimes, for example; this would be an opportune time to run unattended bots or processes for optimal use of the technology infrastructure, Ilan said.

5. Evaluate post-bot performance

There are a variety of reasons that employees don’t always utilize an automated process. Post-bot process mining reveals whether employees are continuing to run manual processes and why. Improving automation use may lead to a quicker realization of ROI.

Process mining after an automation has been established can also identify what Ilan called a “soft” KPI, specifically the tasks employees perform once they’re free of the manual work that has been automated. Typically, those tasks are higher value than the manual task the bot has automated.

Also: Read more about process mining.