Automating manual or inefficient processes is the bread and butter of any technology organization. Here are nine considerations for approaching process automation tasks the right way, and ultimately delivering successfully.
A colleague of mine once quipped that automating a 'crap' process just results in 'automated crap,' and while the language might be a bit uncouth, the sentiment is absolutely correct. It can be tempting when tasked with automating a process to immediately start considering the systems and software to deploy; however, it's worth determining whether the process is currently valid, effective, and necessary before diving into automating it. Furthermore, some processes simply should not be automated.
2. Check your options
Sometimes pure technology is not the right answer for process automation, as the whole offshore process outsourcing business can attest to. While there are myriad risks and considerations to process outsourcing, considering non-technology options for process 'automation' should be part of your evaluation and due diligence process.
3. Look for cheap fixes
Similar to avoiding 'automated crap,' before digging into the technical aspects of automation, consider whether there are low- or zero-cost changes that can make the process more efficient. In particular, look for areas where data are painstakingly gathered but no one knows why they're needed, or complex workflows 'ping-pong' a process between multiple operators when one person could perform several steps. Automation should be icing on the cake of a well-designed process; if you don't have the cake, all the icing in the world won't prepare you for the birthday party.
4. Size the solution to the problem
I was once tasked with building a complex automated order management process for 'selling' scrap paper that had dozens of unique tracking and invoicing requirements. As we contemplated solutions, I asked how much revenue was generated by this process, and was solemnly told it was "as much as $500 annually." While the goal of integrating all sales processes in one system was laudable, a $50,000 solution to a $500 problem was laughable. Simply disposing of the scrap paper was a far more cost-effective solution than expensive automation.
5. Consider the KPIs
When we automate, we usually consider increased speed or reduced cost as the ultimate successful outcome of the automation effort. However, these may not be the right KPIs in some cases. Shortening the time to handle a customer sales call might increase the number of calls you can process, but will reduce sales revenue since reps no longer have time to cross sell, just as a confusing automated system can actually increase customer service calls, producing the exact opposite of the intended outcome. Before reflexively considering cost and speed as your key KPIs, think a level deeper and add KPIs that will mitigate unintended outcomes.
6. Consider the user experience (UX)
Along with well-defined KPIs, User Experience (UX) should also factor into your automation concerns, both from the perspective of the process operator(s) as well as the consumer of the process. Automation that's targeted toward an employee with minimal training will require more robust error handling than one that's in the hands of a skilled operator. Similarly, extra time spent ensuring the process is clearly articulated and displayed toward the operator and end customer will ultimately make the process more efficient by reducing errors and confusion. For example, self-checkout systems at the grocery store were supposed to be a superior option to human-assisted checkout. However, users quickly learned that manually finding and keying codes for produce and bagging their own groceries was inferior and slower than checkouts staffed by humans, resulting in low adoption rates of self-checkout. Obviously, these systems save the store staffing costs, but ultimately they increase customer frustration.
7. Test the edge cases
A key failure point for many automated systems is poor testing. The automation is great 'most of the time' but completely breaks down during an edge case or error condition. While there's a risk of 'over-testing' and becoming obsessively focused on convoluted and unlikely use cases, it's imperative that the automation recover gracefully from predictable errors. What happens if unexpected data are entered? How does the automation respond to a system that's down? How does the automation recover and guide a user when failures occur?
8. Avoid 'orphaned' automation
Like any IT project, once automation is successfully deployed it's tempting to cross the item off the organizational to-do list, and not revisit the automation until it's obsolete or fails. As part of your regular maintenance activities, check the performance of key automated processes. Perhaps there's a new tool or technique that could quickly be adopted for significant benefit, or a minor technology upgrade could provide dramatic returns. Applying continuous improvement practices to your process automation efforts can reap additional benefits from your automation at lower costs than new efforts.
It can be easy to apply the same solutions and techniques to your ongoing automation efforts, but take a moment to consider if there are new technologies or services that might be more effective at accomplishing your automation goals. Emerging technologies like Robotic Process Automation (RPA) or Artificial Intelligence (AI) technologies could be relevant to the automation problem at hand, or perhaps you're tasked with automating a low-risk process that could serve as a test case for advanced tools and techniques.
While process automation is old hat to many IT organizations, to the point that it's seen as a low-level activity that can be given limited attention, there are very real considerations, concerns, and benefits to effective process automation. Executing a difficult technical task often has less visible impact, and associated financial benefit, to shaving some time or improving the quality of a process that's performed thousands of times each day.