The Bot Bother: Thinking Rationally about Automation

by Matt Peters

Many organizations today think of themselves as late-comers to the automation party if they don’t already have a raft of bots working around the clock in the back office.

Like in any new market, there is a time for stepping out and a time for holding back. For those who haven’t yet explored the robotic process automation (RPA) market or those that aren’t sure they’ve got good reason to deploy software bots, the early adopters have reconnoitered and sent word: they work!

But the thing about robotic process automation bots is that they can’t speak for themselves. They are easy to configure, that is true, and, if you configure them correctly, they can speed up and improve laborious and repetitive tasks. But they don’t know what they don’t know. And, if you want them to earn their keep and be productive over the long haul, they need a fair bit of tending. Much like humans – or possibly even more so – you may get a disappointing outcome if you assign them a task and walk away.

Imagine a scenario in which you are building a robotic workforce to take over significant work in your business – invoice processing or data entry – that was consuming the time and energy of many workers. Say your human workforce was able to process 2,000 invoices a day. The RPA vendor claims you’ll be able to process invoices ten times as fast, so you configure your bots and wait for the good news. Here’s the reality: of those 2,000 invoices you’ve been processing, half follow an easy path. They contain the necessary data in the right places and the bots process them easily. Of the remaining half, there is a set of likely exceptions that your invoice expert warned you about, and, if you’re smart, you’ve built logic into the bot to handle that. This increased investment leads to increased success. Then there are the not-insignificant number of invoices that contain some higher degree of nuance. The bot takes a look, can’t intuit what it’s supposed to do, and spits it back. At the end of the day, the report claims only 1,400 invoices were processed – clearly not meeting your expectations.

Here’s another scenario: As you get bots up and running, and you are gratified to know that humans are becoming less involved in the routine tasks and can focus on other activities. Then, unbeknownst to you, an application changes outside of your control – the SaaS vendor makes an update or your own IT team needs to reconfigure an integration. Because the bots are working silently in the background and can’t speak up, they try to access the application but fail. The update has put them out of commission, and there was no plan for humans to jump in and take over the work that had been assigned to the bots. The invoices pile up, and once again you fall short on expectations.

With a user-friendly tool like some of the leading RPA solutions, it’s not hard to see why organizations would be tempted to democratize it. Business units outside of IT are clamoring for it: HR wants it for associate engagement and human capital management. Finance wants it to do contract routing and management. And part of the reason RPA is so attractive is because it’s an equal-opportunity tool that can be implemented to improve productivity or reduce errors in different contexts. IT may find itself saying, “If you want to use it, go for it. We’ll keep the lights on.”

But implications abound. If you program a bot to do the work of a human, you have a different accountability map than you did before. Who do you look to if a robot violates a newly changed law? And how do enterprise systems handle bots operating without coordination? Say your ERP is tuned to handle 300 employees. It has worked well this way for years because no more than 30 people might ever use it at the same time. Now, there are five automation builders working independently of each other in different business units. Each of these new bots – all of which move faster than humans and operate 24 hours a day, seven days a week – happen to access the ERP at the same time, each using it 10 times harder than a single human user. When the system crashes, everyone looks to IT.

Late comers have some clear advantages. They may not be out in front of the pack – and they may need to make up considerable ground – but selecting and deploying bots across the enterprise can happen expeditiously today, especially if the bots have someone who can speak for them. This is IT’s role. Take the reins. Put in place standards for building automations, create support mechanisms and prepare the whole effort to scale. CAI can help you build automation capability and sustainability in the right way and at just the right time.


Want to learn more about how CAI can help you build automation capability and sustainability in the right way and at just the right time? Get in touch.