header-Business Intelligence trifft KI

Retail in Transition: Business Intelligence meets AI

The term “artificial intelligence” is one of the major buzzwords in digitisation. In recent years, technical developments have given AI another significant boost. Intelligent automation and cooperation between man and machine have now become indispensable to the economy. Bitkom and the German Research Center for Artificial Intelligence (DFKI) define AI as follows:

Artificial intelligence describes computer science applications whose goal is to exhibit intelligent behaviour. To achieve this, certain core skills are required in different proportions: Perception, understanding, action and learning. These four core skills represent the greatest possible simplification of a model for modern AI. Perceiving, understanding and acting extend the basic principle of all EDP systems incorporating input, processing and output. The really new thing, however, is learning and understanding part. What today’s “real” AI systems have in common is that they are also trained in the processing component and can thus learn and achieve better results than conventional procedures, which are anyway based only on rigid, clearly defined and firmly programmed sets of rules.

End customers are often unaware that they are coming into contact with AI technologies. Consumers currently witness direct contact with intelligent systems mainly via speech assistants, which are found in most mobile phones and now also in many living rooms. Everything else happens, so to speak, “below the surface”.

“I would say, a lot of the value that we’re getting from machine learning is actually happening beneath the surface. It is things like improved search results. Improved product recommendations for customers. Improved forecasting for inventory management. Literally hundreds of other things beneath the surface”,in the words of Amazon founder Jeff Bezos.

AI: Areas of application in retail

Because of its proximity to customers and the data treasures it has at its disposal, retailing is particularly predestined for AI applications. A survey conducted by management consultants BRP clearly shows how strongly AI will penetrate this industry in the future. Some 45 percent of all retailers plan to use AI to improve the shopping experience within the next three years.

Hertel, Zentes and Schramm-Klein (2011) emphasise in their textbook “Supply Chain Management and Goods Management Systems in Retail” how important the evaluation of data is today:

There are constant changes, they come from inside and outside, and they often come without warning. The ability to react immediately and take back the reins of action, faster even than the competition, distinguishes the market leader from the mere imitator. Action instead of reaction is the requisite. The availability of data is correspondingly important, and especially “real-time” or at least “near-real-time”.

The areas of application for artificial intelligence are manifold. These are characterised by the fact that they would not be achievable without AI, set against a backdrop of excessively large volumes of data. The focus here is primarily on areas of application in the sphere of optimising goods handling processes. In principle, this can be pinned down to five categories: optimisation of promotion campaigns, product range optimisation, shelf space optimisation, consumer insights and price optimisation.

1. Optimisation of promotion campaigns

In retail, promotions and offer periods serve to both generate turnover in the short term and to enhance a company’s profile longer-term. However, the goals here are often contradictory and difficult to reconcile with one another. On the one hand, the short-term attention of the customers is to be won without damaging profit, whilst on the other, valuable customers are also to be retained. The risk with excessive discounts is that customers will be nurtured over the long run to become bargain hunters, who will accept less and less the prices with which a retailer can actually still earn margin. Lastingly aggressive pricing measures can damage the image of valuable brands or even the retailer himself. According to Hertel, Zentes and Schramm-Klein (2011), three dimensions are relevant to planning promotions:

  • Sales receipt profiles: Which receipts contain the advertised items? How many items without price reduction do they list? How high is the residual revenue minus the advertised items?
  • Buyer profile: Which customers feel particularly addressed by the advertised items? Are they customers of a high sales/earnings type, or more so customers who shop frequently? Are they customers you would prefer to retain as loyal customers?
  • Reach: How broad is the customer appeal, or, in other words, how many customers are responding to the advertised items?

The requisite database is often already available in the data vaults of the respective retail company and should be evaluated on the basis of these corresponding key figures. Direct recommendations for planning promotions can only be derived if the relationships between the key figures and the influencing factors, such as seasonality and past promotions, are also taken into account. Intelligent software systems recognise these correlations and determine the appropriate form of campaign from a huge number of potential combinations. This type of promotion optimisation enables you, the retailer, to address your customers in a more targeted manner. Customers who frequently select items from your sports range, for example, should receive appropriate offers in this segment.

2. Product range optimisation

The central tasks to be solved in the area of product range optimisation consist of listing and discontinuation. With the help of intelligent systems, customer shopping carts can be precisely analysed. The influence, for example, that a possible discontinuation can have on the rest of the assortment or, better still, on the customers themselves, can be evaluated. If one of the discontinuation candidates is an item typically purchased by “good customers”, its discontinuation could have a negative impact on the overall outcome.
In addition, intelligent systems can forecast future volumes of goods on the basis of historical sales data and also taking seasonal events into account. These forecasting systems based on previous customer behaviour are also becoming better and better. The retailer often already knows in advance what the customers will be ordering next.

3. Shelf space optimisation

With the help of systems for shelf space optimisation, it is possible to achieve sales promotion for high-yield products in stationary retail through the ideal physical placement of items. Online retail, however, also faces the same problem of presenting the optimal offering in the right place of the online shop. In this case, the recording and analysis of user activity and click behaviour can also provide information about the visibility of certain products.
But despite an optimised positioning on the shelf or the online shop, the following also applies. The customer expects a unique shopping experience, so design your shop for this in an appealing and structured way. For online retailers, we have summarised some valuable tips in our article “Success factors in eCommerce – Some tips towards a profitable online shop“.

4. Consumer insights

A better understanding of the needs and requirements of the customer is decisive to gaining competitive advantages in both stationary retail and eCommerce. Until now, (online) retail has been highly product-oriented. Putting the customer at the centre of considerations, strategies and tactics will mean a clear shift in values for almost all retail companies. Instead of finding customers for products, it is now a question of finding products for customers. So if you know your target group and customers, you will usually not only be offering the right products, but can also perform your marketing measures profitably without any major wastage. The customer also feels addressed and understood, and is thus rewarded with a unique shopping experience.
Too often and also erroneously, customer satisfaction is equated with customer enthusiasm. Their enthusiasm, however, extends way beyond mere satisfaction. Just offering a simple service will not be enough to retain customers. It is crucial to examine and optimise the entire customer journey with the help of intelligent tools – from the past to the present and even on into the future. Only a holistic view, after all, makes it possible to understand the reasons for the entry or departure of customers.

5. Price optimisation

With the help of AI-supported Business Intelligence software such as blackbee, a more tactical price optimisation can be implemented. You can see exactly what price you need to set for a product to stand at the forefront of the market, to increase your profit or to raise your market share.
But even in stationary retail, intelligent technologies can be used to implement differentiated pricing policies that are not simply based on instinct, experience or guesswork. Prices can even be varied at the branch level, whilst taking into account local or regional customer preferences.
Like no other revenue-enhancing measure, price increases have an enormous impact on profitability. Even a one-percent improvement in pricing policy results in an average earnings increase of over nine percent.
Once a strategic pricing process has been established in a retail company, it can then be defined as an improving optimisation cycle. A decisive factor here is the use of analytical processes to interpret the results of those new pricing decisions – the monitoring of profitability and customer satisfaction.

The areas of application for artificial intelligence are manifold, both for stationary and online retail, offering major potential in the spheres of customer loyalty and increased profits. In the second part of this series we will present some concrete practical examples for the use of AI.

Would you like to know more about price optimisation? Contact us now and convince yourself of our technology!