Use cases

Are you able to describe the work? This doesn’t mean your documentation exists or is current. The task could be described by recording a user performing their work on a computer including how they handle exceptions.

Is the work rules-based rather than subjective? Robots need to be prepared (aka, taught, trained, configured) to perform specific actions on your systems. Current technology is insufficient for a robot to determine on its own what to do when faced with a new situation.

Is the work performed electronically? It doesn’t matter how many different applications are required or whether they are in-house, cloud-based, Citrix, desktop or mainframe.

Is the required data structured (or could it be structured)? If not, you may be able to utilize an OCR and/or cognitive application capable of structuring the file.  Alternatively, you could have people enter the data into a structured format. 

Process stability. If your organization keeps changing the process (e.g., responding to competitive factors or new sources of information), then it may not be the right time to automate it. Despite investing resources to stabilize the current activity, you may end up with too much maintenance to keep your automation aligned to business needs.

Target applications suitability. Some applications are harder for robots to use than others. It's a fact that vendors don't really like to highlight in the sales process. Starting with an especially challenging target application could delay the whole program, cause fatigue in leadership and put your credibility at risk. If you have to do it, make sure that you build in an accurate view of the time required.