For a while now, manual tasks have been the main obstacle to increased productivity in many of your compliance and monitoring operations. Yet while the tools at our disposal are getting more and more sophisticated, incorporating them into an existing financial crime operation is still challenging. I am often asked what the best starting point is for integrating automation into an existing work process.
In this discussion, I will review the common steps for successful implementation.
Among the questions I will address are: How do we automate existing processes? Which should be automated, and which should be kept manual? When have we done enough? From my perspective, automation is an ongoing journey, and the first step in the adventure is important to establishing the overall direction and expectations that impact your process steps along the way.
1. Understand the Existing State
In my experience, the best candidates for first automation targets are the short and simple tasks that repeat many times throughout the day. The repetition is what makes those tasks a great starting point. On the one hand, shaving 50 seconds from a 1-minute task can add up when multiple users perform this task up to 20 times a day. On the other hand, user error in those small tasks tends to multiply the impact of that error many times over – copy and paste of the wrong field, clicking the wrong button, can fast track a cascade of other errors.
Those are the kinds of tasks that benefit greatly from automation. But how do find them? I saw many examples of differences between the users’ recollection and the real time “wasters”.
All good decisions are based on data. Automation is no different. We start by measuring an analyzing what the users spend their time on. “Over the shoulder” reviews with a notebook are a great starting point. Various recording tools are another good way which to allow many people to review the collected information. At NICE Actimize, we started using the Automation Finder tool for machine-learning based analysis and identification of candidate tasks, but any approach works.
2. Pick the Right Tools
Once we know on what tasks our users spend their time, we can find the best way to automate it. In most cases, there are several ways to do this. Not all of them give the same results over time. Each tool usually provides several integration options, and a multitude of choices from which to pick. There are pros and cons to each one. While back-end or “API” integration provides more robustness, front-end or “cursor driving” integration is often much quicker to implement. By analyzing and comparing the different options, it’s also easier to identify the mandatory requirements. For us, these are usually flexibility and auditing. Flexibility refers to the ability to easily change and deploy to production updates to the automation methods as the tasks or connected systems are changing. Auditing any user decision or action by the tool is a mandatory requirement to make sure we can understand how and why things happened.
3. Keep Track of Your Choices
The most common reason for failure of automation projects is picking the wrong target process. There are many reasons why a specific project might be the wrong one. Complexity, external dependencies, rapid data changes and other factors can all make a project much harder. It’s important to remember that the automation journey is not a one-time investment.
As we successfully automate more and more processes, we gain better understanding of what works and what doesn’t for our organization. This analysis should be done throughout the project and not just after completion. Identifying potential issues early provides more time to address them. One of the results of that analysis can be a decision to shift the focus of the automation. Many times, as we dig deeper into the target task, we identify additional “sub-tasks” in the process. Those might be optional steps or preparatory steps that take a lot of time to the user. Automating them might be a bigger win than the basic task!
4. Rinse and Repeat
The mindset of successful automation is of a rapid iteration process. We already know that over time, automated processes need upkeep. Either to adjust to changing business needs or because the connected systems have changed. Therefore, the earlier we can get them to production, the more time they’ll stay current.
Unlike many other systems, each automated process has an immediate benefit – one less task for the user to do. As our users start using the automation tools, it becomes easier to identify and prioritize the manual tasks. An agile design and deployment process allow us to take advantage of this and provides the most benefits to our users.
5. Create Reports
I’ve said before that the best decisions are data driven. Collecting and understanding that data must be a basic goal from the beginning of any automation process. For most automation projects, that means reporting on the usage and utility of each process. The resulting reports identify problems and improve the continued development of new automations. People stop using an automated process is if it stopped working as expected. They also stop using it if the business needs have changed. Either case requires us to understand the cause and update the automation. Making sure that all automatic processes are up to date and valid is critical for the continued faith of the users in them.
I see automation as a long-term initiative. There are always places to improve or extra tasks and processes to automate. But like any journey, the most important step is the first one. The best results come from a thoughtful design and deployment process. This is then combined with agile iteration and good reporting. I’ve seen the impact experience can have on the process, and it strengthens my conviction: now is the best time to review and automate all parts of the operational process.
What challenges did you encounter when deploying automation tools? Drop us a note at firstname.lastname@example.org to share your thoughts.