Injecting Robotic Process Automation to augment value in Master Data Management
Introduction
“Practice makes a man perfect”, a proverb we all have heard and grown up with, however, human error still percolates in the mundane day to day activities. For instance, forgetting car keys, spilling coffee because you are in a hurry etc. On the other hand, our alarm clock never fails to ring in the morning. This leads us to a long-debated topic of human vs. machine. With each technological leap forward, there is a parallel rise in fear that humanity will somehow be displaced. However, Ben Jones, Google’s creative director says, we need to stop thinking of machines as rivals and instead should see them as opportunities. We all would agree that in repetitive tasks — a machine is far more efficient than humans because it never forgets the rules. By 2023, there will be a 30% increase in the use of RPA for front-office functions (sales and customer experience), according to Gartner, also by 2024, the organizations will lower operational costs by 30% by combining hyper automation technologies with redesigned operational processes. Robotic process Automation (RPA) has matured over the past few years by automating back office processes focused around Finance & Accounting (F&A), Customer Services, Human Resource Management (HR). This article focusses on how RPA can enhance value in Master Data Management (MDM) by digitally emulating the manual activities of MDM users to accelerate process automation.
Rise of Robotic Process automation
Since inception humans have been trying to simplify daily chores to work efficiently. Similarly, organizations have been automating the repetitive tasks so that people can focus on more complex tasks. Figure-1 gives a graphical representation on how automation in organizations has evolved over time, giving workforce enough time to focus on high value tasks.
Business Process Management (BPM) has been playing a significant role across organizations to streamline the existing processes to increase efficiency. Despite the numerous advantages that a BPM solution adds, there remains still a large part which demands manual effort and thus, is prone to errors as well. RPA on the other hand follows a bottom-up approach and optimizes the process by bridging the process gaps and automating it. Before, we dive into how RPAs can enhance MDM processes, lets first understand a typical RPA.
RPA is an application of technology, governed by business logic and structured inputs, aimed at automating business processes. Using RPA tools, a company can configure software, or a “robot,” to capture and interpret applications for processing a transaction, manipulating data, triggering responses and communicating with other digital systems. RPA scenarios range from something as simple as generating an automatic response to an email to deploying thousands of bots, each programmed to automate jobs in an ERP system.
Majority of candidates for implementing RPA solution can be found in back office areas. Below are few examples of the processes in back offices where RPA is widely used.
Back offices areas often face challenges performing repetitive, rule-based, redundant and manual tasks. Typical steps involved in the process –
· (COPY) Search, Collate or Reconcile data from disparate systems/applications/screens.
· (DECIDE) Repetitive and Rule based decision making processes.
· (PASTE) Input the corroborated data.
· (EXECUTE) Run the application.
After a use case is identified based on ROI considering the factors such as: frequency, complexity and repetitiveness of a task, a bot can be deployed with ease as-
· RPAs are nonintrusive packages that do not require companies to make changes to their existing technology stack but reside above the enterprise applications.
· Basic RPAs can work with rule-based processes and leveraging state of the art AI/ML they can work within all cognitive periphery.
· Bots can be scaled within a short time period to handle the increase in workload.
· Require less investment upfront and can be executed in agile fashion
· Based on the level of supervision required in a task, a bot can be easily integrated in the existing process as attended or unattended.
Leveraging RPA in Master Data Management (MDM)
Accurate and complete master data provides a single authoritative view of information for better decision making. However, most organizations find it challenging to create and maintain it.
Poor data quality is also hitting organizations where it hurts — to the tune of $15 million as the average annual financial cost in 2017, according to Gartner’s Data Quality Market Survey. Automation can negate human errors in the operational processes within an MDM and will result in improved data quality, higher process efficiency and compliance to regulations at a lower cost.
For instance, in retail industry, bringing a new article to market in-time is always a challenge for the retailer. Manufacturers introduce new products frequently in the market but for the retailer to procure and sell that article, master data needs to be set-up and the product-store listings should be complete. Figure-3 below shows a basic process flow for a retail product from procurement to sale.
Article details might be shared via e-mails in different formats such as spreadsheets, text or word documents etc., might involve multiple follow-ups from different stakeholders, and then gets approved, leading to master record creation. Despite strict validations from MDM teams, there are strong possibilities that the master data record created will not be accurate. The manual task of repeatedly validating details sent by supplier, searching & collating company-specific information from disparate sources or multiple stakeholders, is monotonous and is a potential risk on data quality.
Figure-4 displays a completely manual process. This manual process can, however, be completely transformed by unattended bot that executes under no supervision, and thus automating this process. Figure-5 shows how an unattended bot can read article data from an e-mail attachment or might use AI/ML to extract data from unstructured documents. Rules are set-up for the expected scenarios based on the inputs from business and MDM users. Bot validates the article data shared by supplier and populates the company specific or custom fields based on the configured rules. After the necessary data needed to onboard the article, bot creates an article master by calling an API or a function module. The supplier is notified through an email and the request is closed. The bot activities are logged for later analysis and troubleshooting. An attended bot could also be deployed if the MDM team wants to verify the final master data before creation
Organizations can argue that a strategic long-term solution could be to leverage an onboarding portal in which suppliers could submit article details for creation or aim to integrate the MDM/ERP with the Global Data Synchronization Pool (GDSN). However, even with this seamless integration and structured workflows, the MDM team will still need to reference some data points from different sources before creating a record. The data referenced here will be company specific and could not be provided by the supplier (Figure-6). Additionally, updating the custom fields in an MDM system will always be a challenge and would need manual intervention from the user. Every company customizes the standard MDM/ERP systems to suit their business requirements — custom modules to support POS systems, Offer management, Recipe Management for BTO products, custom hierarchies etc. Therefore, deploying a simple unattended bot could completely automate the tedious onboarding process and drastically improve the time-to-market for an article. Tactically, bots can be deployed at a lower cost to automate the manual process till the onboarding portal is being developed or the integration with GDSN pool is in progress. This can help to ease the load on MDM team and create a more efficient solution.
Sometimes the logic for selecting data for custom fields cannot be defined based on a single decision tree, which makes it difficult to follow the rule-based approach. An ensemble of decision tree can be leveraged as a ML service to predict the input data for article creation (Figure-7). By combining advanced analytics capabilities with process automation, an advanced form of RPA can be designed — a bot that can analyze, comprehend, and draw conclusions that are better than a human rationale, from both structured and unstructured data. The model will be vigorously tuned with each wrong decision, making it robust to use across the enterprise.
Conclusion
This article discusses a simple onboarding use case for a retail industry. However, there are variety of use cases for MDM in which RPA can be utilized to maintain master data for an organization — such as customer/supplier onboarding, site master maintenance, updating master data record based on triggers etc. For unstructured data, the bots can leverage OCR (Optical Character Recognition), pattern recognition, NLP to extract relevant data from various formats and automate parts of manual data entry. Additionally, RPA could also be leveraged as a tactical solution to optimize the process till it is standardized.
With right motivation, approach and design, a complex process that has repetitive tasks and refers to specific rules can be easily automated. The productivity level of workforce can bolster, and they can focus on high value tasks rather than performing mundane and manual activities.