When implementing an RPA solution within your organization, what did you find to be the easiest process to automate, and why?
Were you expecting the process to be easy? If not, what made it easy?
Do you think the process would have been easy with any vendor or only the vendor who you implemented with?
The easiest to automate was to download a report from one system and upload it to another. It was easy since both systems are simple and stable.
I was not expecting it to be that easy, however both systems have great performance and they react to the default speed of our Kryon solution without the need to fine-tune it.
-Easy to data entry process web scraper
-Reading pdf
-String manipulation
-Mail reading
-Excel process
-Use cases in websites.
-Data Entry
-Email reading
-Data monitoring for payments
-Shipment status
What made it simple is rule base, low number of application, no Citrix, no handwritten, no exceptions.
Based on my experience, the easiness comes from the type of processes itself and not from the vendor.
However, the vendor with builtin AI like ML, Computer vision ,NLP can deliver faster medium and complex processes.
We have implemented unattended robotic process automation.
We found the following factors affecting the automation complexity:
1. Number of operational systems involved in the process
2. Number of required transactions
3. Number of stages in the flow
4. The number of branches in the process – conditions that cause the flow to split.
5. The complexity of exceptions handling.
6. Quality of cooperation with the involved business partners.
When we chose the processes, we considered 2 argument: complexity of processes and their financial benefits.
If we talk about complexity of processes, so we considered:
1) Number of system
2) Number of actions (screen forms)
3) Number of decision
4) Number of parts of the process that are fully digitized
If we talk about financial benefits, so we considered:
1) Number of transactions
2) Duration of transactions
And then we chose the process with a heavy workload (Number of transactions*Duration of transactions) and low complexity (at first number of parts of the process that are fully digitized)
For example, creating a sales order or creating reports
1. Process that is already 100% electronic (no paper)
2.Process that is standardized
3. Process that is not expected to change in the near future
4. Process follows a workflow (defined steps to execute)
As per my experience in process automation. I have come across easy and difficult processes to automate.
When I say easy, it all depends on how stable the application you are automating is, understanding the requirements, and implementing all cases. Also, the best practices that I followed made processes easy to automate. Difficult processes that I came across where there was a change in requirements and application screen constantly made it a bit difficult to automate.
When implementing an RPA solution within your organization, what did you find to be the easiest process to automate, and why?
Answer: Process which is very routine and not many patterns in the input data. This would create a positive impact within the organization which will help to automate more processes.
Were you expecting the process to be easy? If not, what made it easy?
Answer: Not necessary. If there is an opportunity, we will simplify the process.
Tasks that are repetitive are the ones to focus on.
RPA as a tool has evolved significantly in recent years to become more of a “intelligent strategic” automation tool compared to its predecessors which were more inclined towards tactical automation.
Process selection is typically based on enterprise maturity and the skillset at hand.
If organization has already evaluated RPA and now wants to roll out automation across its departments, then the first criteria is to do a detailed process discovery to understand the classification of the processes (core vs non-core). Noncore processes are a better suited candidate to start with as the risk associated is low.
If the organization is starting on RPA journey and wants to decide on the type of processes that should be used for RPA then it should consider processes which are linear in nature and has a fixed set of steps, low is exceptions and are well controlled, provide immediate benefit to a human resource, are standardized in terms of data exchanges, are not required to dynamically generate or consume data et are a good candidate.
Criteria listed in the second bullet are typically considered in what we call as a jump-start engagement. Post the jump-start, canvas widens to consider a lot more variables for selection. The below blog provides good information on selection for jump-start.
www.prolifics.com/blog/rpa-getting-your-jump-start-right
To control the implementation, key is to control the requirement and hence the design. If the design is stable, then the implementation can be controlled. Hence as a start, our every RPA engagement utilizes an automation analyst who ensures that the models are well crafted before developers fly-in. This is key to ensure a successful project.
As I belong to a service provider, so I will like to refrain from “Do you think the process would have been easy with any vendor or only the vendor who you implemented with?”.
The easiest processes to automate were the manual processes requiring no human judgement, with few dependent systems like legacy systems. We were somewhat expecting the process to be easy. Access to few or null systems made it easier as well as the logic of the automation. I don't think the process would have been easy with another or any vendor. Vendors make simpler and easier implementations more difficult.
It's important to include Kryon on your radar. That is the one I work with today.