In today’s world, every action we perform generates data. Similarly, the business processes we focus on today for automation deal with huge amounts of data. The business processes receive data from different sources such as:
- Different internal source applications
- External applications
- Scanned, computer-generated documents.
- Social media and many more…
The data processed within the business processes are sometimes well structured. Still, in some scenarios, the data are very unstructured depending on the source and the nature of the business process.
Further, the Business Intelligence and Data Science teams in the organization perform many manipulations, transformations, and analytics on top of this data to generate various kinds of reports and dashboards. Process managers and top-level managers use these reports/ dashboards to carry out strategic and operational decision-making. Additionally, such analytical data is also used in downstream applications to perform different functions based on the analyzed data.
Even today, many high volume and repetitive processes are still performed manually by business users to carry out the daily work. On the other hand, many data scientists, data engineers, and BI engineers work relentlessly analyzing data, cleansing, and building complex models to generate predictive and other analytical data for decision-making and downstream applications in the process. Robotic Process Automation (RPA) enables easy automation of repetitive functions. However, operations that require forecasting, geospatial, or predictive data to perform specific actions were not fully automated because data scientists manually performed data analytics-related actions before passing them to downstream subprocesses. This disconnect between RPA and analytical data generation restricted automating such processes end-to-end.
Why data analytics from personal experience
I started my career eleven years back as a data analyst. Since then, I have been in the data analytics domain for many years. I have experienced the importance of extracting information from different silos of data and generating analytical information to support decision-makers and operational activities. Generating analytical data is a very painstaking activity due to some of the following reasons:
- Different sources have their distinct ways of storing and representing data.
- Data used in business processes are sometimes structured and sometimes unstructured.
- Cleansing and transforming data into meaningful and usable format takes a lot of effort and time.
- Data quality in source systems may be low due to human errors and limitations of source systems themselves.
There are many tools out there that enable data scientists and data engineers to extract, transform and build meaningful analytics out of the raw data.
Vendors that provide analytical solutions:
- Oracle, and there are many more…
Based on my experience, building and using analytical systems are done separately from the primary process as it requires special skills.
RPA + Data analytics
Many vendors provide RPA solutions. However, UiPath provides the best state-of-the-art RPA platform to automate manual, repetitive actions performed by humans. UiPath robots can easily interact with any software platform to mimic human interactions using multiple technologies. On the other hand, many Analytical Process Automation (APA) tools such as Alteryx provides the capability to easily build a dataflow to perform various complex data related actions such as:
- pulling databases, files, and API’s
- Prepare cleanse and combine data sets
- Perform complex forecasting, geospatial, and predictive analytics
- Send data to files, databases, APIs, BI/ reporting/ dashboard tools and, many more.
By just looking at the points, you might wonder, “isn’t the RPA tools capable of doing these?”
The answer is:
Yes, the RPA tools can extract data from different sources, cleansing and sending data back to files and databases. However, performing forecasting and complex analytics based on large amounts of structured and unstructured data requires a lot of cleansing, transformation, building specialized predictive models, and many more complex actions. The goal of an RPA is not to build analytical models. Hence, there are specialized tools such as Alteryx to perform such activities.
While RPA and APA provide robust solutions on their respective domains, I would like to take this conversation around how powerful the combination could be for automating end-to-end processes.
What makes this combination so powerful?
RPA is capable of automating high-volume repetitive business processes by introducing a virtual workforce. The automation is developed based on the business requirements and the outcome of the data they expect. Hence, you can see that; RPA or APA, it is all depending on the data. Many business processes in industries such as finance, banking, logistics consist of steps that require the generation of predictive, geospatial, and other types of analytical data to perform critical procedures.
For example, in the supply chain process, examining the demand for products based on the historical data and calculating when to order which product and by what quantity is a critical step to maintain the consistent supply of goods. RPA could assist in extracting all the required data from multiple source applications. Further, RPA can also help update internal applications and perform the purchasing of stocks automatically based on the data we feed. However, a team of experts must sit and manually perform the calculations to come up with the predicted stocks to order. Performing such predictive calculations manually by humans would require an immense amount of effort and time, which may drastically decrease efficiency. In this particular scenario, if the analytics are done manually by humans, it would not provide the expected efficiency rates by automating. However, introducing APA could generate the predictive data within minutes, which increases the efficiency drastically. Such analytical power combined with RPA workflows enables the RPA robots to harness the power of APA by communicating with the AI and ML models to retrieve the data it needs to automate purchasing of goods.
The following image illustrates the same process discussed in a graphical way to show how APA fills the gap for data analytics by communicating with RPA robots.
Is this concept possible?
Yes. As we speak, UiPath, the worlds leading RPA platform, offers the capability to connect with Alteryx, one of the world's best APA tools to perform data analytics. Alteryx is an APA tool with a very high demand worldwide to automate analytical processes in different industries such as banking and finance, healthcare, retail, education, sales, operations, etc.
How to Approach?
It all starts from the requirement gathering phase.
Coming back to basics, just like for any RPA project, a thorough discovery is required to properly understand and capture the goals and reasons of the process. RPA is not about automating the same old process as it is. As experts, we provide solutions to automate the process while finding ways to optimize and standardize the process by applying process standardization and optimization methods.
Considering the points mentioned below is very much essential during the process discovery state.
- Challenge the current process
- Identify all the different data sources of the current process.
- Capture the exact steps carried out during the process
- Capture at what point in the process it requires to perform the analysis of the data
- Capture the types of analysis performed on the raw data and how to streamline the steps
- Capture what is the final output of the analysis process
- Capture how the generated analytical data is used in the downstream requirements.
- Identify what other manual processes use the generated data and the steps carried out in those processes.
- Identify the process inefficiencies and data issues that need attention.
Once all the requirements are captured and analyzed, start laying down your solution. As an additional step, explore how RPA robots can easily communicate with APA tools to generate the analytical data based on the data extracted by RPA processes and wise versa. Building a solution that communicates between the two platforms could reduce manual effort and increase efficiency and reliability. The combination also enables to automate processes that were previously not able to automate end-to-end.
Every business process is unique. It is vital to properly understand the exact requirement for RPA and APA before committing to any technology or concept. Combining RPA and data analytics expands the scope of RPA, enabling us to consider more and more business processes for automation. Considering the facts, the combination of RPA and APA could actually be a game-changer for any organization.
I will also publish a demo on how UiPath works with Alteryx automating processes end-to-end soon…