Gartner defined algorithmic merchandising optimization as “an enabler for retailers to more precisely determine items that need to be displayed and stocked, as well as how they should be priced and promoted, to maximize sales, margin, inventory, and customer satisfaction across touchpoints.”
Post the disruptions in the past year due to the pandemic, retailers were forced to reassess their business operations, resulting in a significant boost towards embracing digital transformation at scale. Retailers are beginning to adopt technologies with a two-fold focus – attracting and retaining customers, and optimizing costs, thus accentuating the importance of technology to the industry.
Algorithmic retailing combines the power of artificial intelligence and advanced analytics to transform retail. Several advanced analytics technologies are already used by retailers, who are also beginning to leverage machine learning algorithms, smart data discovery, context-aware computing, and deep learning technologies.
There is a drastic shift in the retailers’ mindset with regard to the decision-making process. Algorithmic merchandising, on the back of this trend, is experiencing increased adoption from retailers, who are focusing on building business resilience while being customer-centric.
Gartner’s Hype Cycle for Retail Technologies, 2021, states that “Algorithmic retailing connects big data to results, navigating a journey from descriptive to prescriptive analytics. This journey includes the identification of data sources, use of automation and advanced analytics, and application of algorithms and artificial intelligence that will lead to highly repeatable and tenable business processes. It is the use of mathematical algorithms, data discovery, advanced analytic capabilities, and AI, combined with automation, to drive effective decision making.”
Merchandising enables retailers to attain higher sales and margins, forming a key part of algorithmic retailing. Complex analytics is seamlessly integrated into the technology, helping maintain a customer-centric approach for smarter decisions at any level of the retail organization.
We shall explore the impact of algorithmic merchandising at different stages of a seasonal or cyclical retail business below.
This is a crucial planning stage where retailers forecast demands helping drive allocation, production, and sales planning. The analytics systems focus on predicting how products will fare in the market. As there are multiple other variables impacting the demand, retailers do not completely rely on data-driven demand forecasting models. Even though analytical tools are used, a lot of demand forecast is generated through instinct and experience-based micro decisions. For example, in fashion retail, new product demands are based on multiple factors like style, fabric, color, cuts, patterns, fit, finish, and texture which eventually influence shopping decisions.
In such instances, there is greater control over the use of these variables due to AI and machine-powered analytics systems. This results in higher accuracy of demand forecasts. Business context, real-time data, external influencers like weather promotions, social media reviews, the performance of similar products, etc. are taken into consideration by AI-augmented analytics to forecast demand.
On the other hand, accurate demand forecasting determines inventory management, pricing decisions, customer engagement strategies, and more.
The focus of this phase is on the execution of the sales plan to generate maximum profit from the inventory. Product sales across stores and inventory movements need to be closely monitored and controlled.
Anomalies result in stock out or stock excess, which need to be cleared off with heavy markdown at the end of the season.
Businesses today rely on algorithmic processes as thousands of products and styles move across hundreds of stores. These processes can detect patterns and draw attention to products that need a temporary price reduction early in the season to maximize revenue for the remaining season.
These anomalies help expose other business opportunities like altering store merchandising or running behaviorally targeted campaigns to boost sales. Retailers in this phase leverage algorithmic retailing for inventory optimization, price recommendation, assortment tuning, new product sales optimization, customer-centric offer recommendations, personalized promotions, and many others, resulting in optimized sales and profit.
Now, the products that have not sold adequately despite the interim season optimization efforts have to be marked down in a particular promotional window for clearance. Artificial Intelligence and machine learning techniques are useful in processing answers to questions like – “What percentage of markdown price is optimum for each product to clear off its inventory?” Or “What kind of products have the most chances of a sale in which store locations?”
Algorithmic markdown analytics are used to continuously modify and shape markdown prices for the highest return on inventory in each store cluster and store location. It analyses the products that need to be discounted, the amount of the discount, the price elasticity, competition from other retailers, ongoing promotions, bundling with other items, other marketing techniques, shelf placements, and so on.
AI algorithms come up with various decision trees at the same time on a variety of sub-groups and then combine them all to present a predictive solution. They can also interface with pricing systems to automatically implement recommended price changes across the store network, thereby making it easier to implement the pricing decisions.
Thus, we see that an upturn in digital business opportunities has forced retailers to manage complex opportunities that exist for the smallest of moments by relying on algorithms. In addition, complex decisions like product pricing, merchandise sorting and distribution, and marketing messages are made by smart machines with the help of algorithms.
Multichannel retailers are continuously seeking improved ways to drive decisions through advanced analytics. Although few of the algorithms will be ubiquitous and shared for the benefit of the industry, there will be algorithms directly aligned with the interests of the retailers and needed to be kept confidential. With the growing number of algorithms, there is an expansion in the quality variations as the algorithmic economy comes to life. Hence, retailers need to figure out a way to manage algorithms and oversee which and how certain algorithms can be shared, and which ones can help provide a short-term competitive advantage.
