Many retail organisations have already taken a data-driven, automation-oriented approach to address a range of business issues but they are also poised to take advantage of artificial intelligence (AI) and machine learning models to help solve more complex business problems.
For some, Robotic Process Automation (RPA) has been producing benefits for over a decade, helping to automate manual, time-consuming processes and eliminating the need for humans to perform repetitive, transactional tasks like gathering and sorting data.
However the real challenge is how to unleash RPA so that it can overcome key stumbling blocks to aid digital transformation strategies. Valuable data is often unstructured, embedded in documents and emails, and cannot be processed by a simple RPA solution alone. RPA can only automate simple tasks. It needs processes to follow finite predefined rules with structured data.
The answer to this conundrum lies in the convergence of AI and ML with RPA to create Intelligent Automation (IA), which enables AI/ML to automate the prediction decision-making and RPA to proceed with the manual next steps within the process, without human intervention. In other words, adding AI/ML enables the RPA to ‘think’ through more complex problems.
It’s this ability to connect the head with the hands that offers the potential to increase the range of knowledge work that may have previously been considered to be either too complex to automate or required human intervention to make predictions.
This is achieved by firstly training models on historical data to make predictions. This involves collecting and preparing the data - often the most time-consuming step in machine learning - and concluding with a training data set that is labeled and ready for modeling. Models are then built using algorithms to address different types of data problems, i.e. classification, regression, binary. Secondly, once the model is built and deployed into production, the next phase of machine learning can begin where unseen data is scored against the built models. This is the step where RPA can ask the machine learning model what to do next, with the model providing a prediction decision for RPA to continue without human intervention.
For those in the retail sector, combining AI/ML with RPA can add real value across the entire ecosystem. It can help the industry to create a scalable digital workforce that has the capacity to execute processes that don’t require human intervention and deliver a return on investment in less than 12 months.
Releasing human labour from mundane tasks for higher-level, human-led decision making can, for example, contribute to far greater accuracy in demand forecasting predictions. Improving efficiency throughout this process can be achieved by using historical data to predict demand for each SKU/Store and leveraging RPA to order the correct amount of supplies needed in the ERP to improve less under-forecasting (improved product availability) and less over-forecasting (reducing excess stock). This is a great example of RPA and machine learning working together to drive better outcomes.
Providing greater visibility of data in real-time is also being used by retail companies to predict the probability of returns for items purchased through all channels. This helps minimise risk of excessive inventory (reducing inventory holding cost and working capital in the supply chain) as well as reduce trans-shipment cost. By predicting how much stock will be returned, less stock will need to be procured from suppliers, retail companies can operate more effective return policies and incentivise customers to not over-order goods.
Intelligent Automation is being harnessed to optimise promotional campaigns to identify the best SKUs and best promotion strategies (e.g. rebate, discount, BOGO, etc.) to achieve target revenue or volume. Using an historical understanding of the relationship between price, promotion and demand, retail companies can plan for different promotion strategies and execute the delivery of those plans with RPA. This will increase sales revenues, gross margins, product sell-through on promoted items, and vendor support.
The ability to standardise data, use larger data sets, remove biases, and train algorithms more efficiently to identify, for example, which customers are likely to complain in the next 30 days, can provide nightly predictions based on customer behaviour to identify the most at-risk customers, thereby reducing customer churn and increasing profit margins.
Introducing more automation in retail warehousing will also enable data to be linked back into manufacturing, and other data lakes, to provide greater visibility of trends, faster and delivering scale manufacturing, and more agile supply chains which are major requirements, especially at this time.
Identifying the most effective product to offer each customer can influence their buying decisions and historical data (purchases, web searches, etc.) can be used to predict the propensity for customers to respond to different offers. Delivering personalised offers to customers and collaborating with category managers to optimise the best offers and offer types will create an increased response rate and decrease the overall marketing cycle times. This will encourage customers to shop for new categories and increase category penetration for retail.
Intelligent Automation is also enabling the retail industry to manage and integrate legacy systems and achieve the benefits of digital transformation without updating software, developing APIs, or building a new system, within weeks, rather than months or in some cases years.
Data can be collected from multiple sources and must be cleansed and prepped before modelling is initiated. However instead of being locked in an ivory tower, Intelligent Automation is democratising AI and RPA, providing people with direct access to data science so they can make use of the information themselves. There’s no need to wait to gain access to the same information from a group that is siloed somewhere else.
Enabling the retail industry to take advantage of these AI, ML and RPA tools and techniques to support AI-driven decision making and deliver ROI in a short period of time is increasingly becoming a practical reality. Organisations are already solving data-driven machine learning use cases such as daily demand forecasting, customer loyalty state, predict next best offer, and vendor invoice fraud and that’s just the beginning.