Hello there! Book a 30 minute demo.
Get started

The chicken and egg problem of automation

Which came first, the problem or the solution?

If asking Edward Deming, the answer is clear:

“It is not enough to do your best; you must know what to do and then do your best.”

However, the competitive landscape is so tight that many companies just don’t have enough time to analyze what to do before jumping to the solutions for quick digital transformation. We buy platforms and then try to find enough good use for them. And if that is not enough, we buy even more platforms, combine them, and call it hyperautomation ;)

Could it be that if knowing the problem better, the solution could be more straightforward and needing actually simpler automation? Let’s find out.

How do companies make automation discoveries currently?

  1. The traditional manual way

That is something we all know. We buy automation capabilities first and then use best-practice frameworks, browse use-case libraries, interview people, and collect ideas from employees. 

  1. With the help of technology - the usual suspects

There are two mainstream ways of trying to solve the automation discovery the digital way. The RPA market boom has boosted both. Visualizing event logs of the business applications and/or tracking keystrokes and mouse click of process specialists. Short intros below if you’re new to those.

Visualization of event logs to understand the process, e.g., process mining

Process mining was originally created for end-to-end business process analysis, design, and enactment (triggering further operations). It is based on event logs to bring visibility into processes within enterprise systems, like ERP and CRM. It is an excellent tool for analyzing and optimizing significant processes with large transaction volumes and business criticality.

What are the process mining requirements for automation discovery?
  • Business systems need to be integrated, also the outsourced and custom ones without integration templates.
  • Data quality must be on a sufficient level to support discovery.
  • Work needs to happen on event logs, not documents and desktop apps.

Example: Celonis is the leading company for large end-to-end process development giving significant savings for companies when even small changes have a big impact.

Keystroke analysis to understand the tasks in detail, e.g., task mining

In contrast to system data and process mining, task mining gives an understanding of how the subject matter expert does the task in detail. When a potential use case for task automation is already identified, this approach is good, helping to create specifications for automation documentation.

Example: UiPath Task Mining and Capture create specifications for their RPA robots. That saves time when you already know that RPA is the right solution.

Bonus: Task mining is a nice evolution from the Windows Task Recorder that I used to build my first RPA specifications in 2015. And it comes free with Windows: Create documentation or training with Task Recorder - Finance & Operations | Dynamics 365 | Microsoft Docs

What is needed to understand before jumping into the automation solution?

The answer is — the problem itself. It’s about understanding what we’re willing to automate; system, process, or tasks. How do the solutions mentioned above fit these problems:

The manual approach works well for all those if there is enough time and knowledge to drill into the root causes and go through all the options. Typically we don’t because time is expensive, and going through all the possibilities for every stakeholder is also slow. It is also hard to get time from subject matter experts who know the processes in needed detail. Where does this lead us? We try to save time and focus on the easy-to-spot problems that are only the tip of the iceberg, resulting in automating lots of tasks instead of understanding the root cause. Mr. Deming is not happy. 

Would the technology make Mr. Deming happy then?

Task mining is good for understanding task-related problems in detail. The problem is that you cannot do digital transformation through task automation. It does not help improving processes and systems. E.g., if the problem is a lack of data integration, doing copy-paste faster is only a band-aid. It sounds like scaling the root cause of problems.

Process mining is good for understanding documented process-related end-to-end problems. It’s suitable for process harmonization and ensuring that everything goes as planned. Regarding digital transformation, the problem is that the main flows in core systems are already automated. Typically we’re more interested in what happens outside of automated processes.

Where Workfellow stands in the field then? We’re like process mining for knowledge work to understand the underlying system problems first and then understand the manual bottlenecks in processes and tasks - all this automatically. This way, you can choose the best solution for the problem: let it be process optimization, task elimination, API, low-code, settings, way of working, productivity apps, B2B SaaS, integration, RPA…

Workfellow clustering work

Workfellow is not based on event logs, so it does not require integrations, configurations, or historical data. We’ve designed it to serve the full spectrum of development people: process owners, leaders, automation teams, and IT. The quick wins are on us: check our demo.

Written by

Henri Wiik

CPO & Co-founder at Workfellow. 10 years of experience in Automation & Digital Transformation.