There are no brick-and-mortar self-service specialist shops. Yet most of on-line shops work that way. How to leverage the knowledge of experienced salespeople to better serve e-customers? AI to the rescue!

In a traditional brick-and-mortar shop customers get personalized service. A shop assistant is there to offer help, answer questions, recommend or point to a certain shelf or product.

A self service shop is different – shelves are arranged in a way characteristic for this shop, and the shoppers are left on their own. On subsequent visits to the shop they memorise the layout. On one hand it speeds up the shopping process, on the other it bonds those customers to this brand. This predictability makes us feel safe, and that makes us come back to shops we know, not look elsewhere. An on-line store is in a way like a small self-service shop, with categories instead of shelves and virtual shopping carts.


Has anyone seen a self-service specialist shop? Probably not – and for a couple of reasons. First of all, customers of specialists shops come for advice, not just to shop. Secondly, product details play a lot more important role there, and the fact that there are many similar products to choose from, does not help make the final choice. It takes either specialist knowledge or true dedication to go through all the specs and chose the right thing. It is so much easier to just ask the shop assistant.

Of course there are specialist shops within the on-line realm. But do they make enough profit? For sure they struggle with the aforementioned issues, issues that render self-service specialist shops impractical. Lack of, or limited assistance to the buyer means they are left with comparing details and finding answers to their doubts on their own. Anyone who has ever been faced with such task knows how tedious an informed choice can be.

pig in a poke

Even if customers makes use of the offered filters to find the perfect product, they are often confronted with an empty list and a “There are no products matching these criteria” message. OK, so much for the perfect product, they might think, and turn back to filters – but which? Or, instead, they might go for the pig in the poke approach, shouldering the risk of a mistaken purchase and returning the product.

Obstacles en route to optimum choice often mean the purchase is not finalised or that the customer would never return. What can be done to change that? “Hire a knowledgeable shop assistant who would help and advise customers!” you might say. Such solution, though simple, is costly, and in the 24/7 world quite impossible.

Assistant or AI?

So, if shop assistants are expensive and need work/life balance, how about harnessing the power of Artificial Intelligence (AI)? Surely it must be effective, with all the recent hype. True, but we must bear in mind its limitations – we want intelligence, and one that is not too artificial. AI to function properly needs a lot of examples: tens, if not hundreds of thousands of examples to learn from, to build reliable patterns of behaviour. When data to feed this AI is scarce, resulting recommendations become appallingly irrational.

Is there a way then, to combine the advantages of both approaches while eliminating their apparent shortages? The answer is yes. FeeCOMPASS® utilises the collective knowledge of experienced shop assistants and advisers, amplifying it with the help of AI; feeCOMPASS® is there, 24/7, to help your eCommerce customers make the best choice.

AI ? HI, or when the whole is greater than the sum of its parts

In the machine learning process, AI reads and systematises the knowledge acquired in this way. It estimates its ignorance by determining the probability with which its answers are true. It is exponentially proportional to the volume of text used for learning. Unlike in machine learning, in feeCOMPASS, the source of knowledge is human and it is an elusive knowledge, located in the heads of salesmen, resulting from their experience. They share this knowledge with the system, which in return quickly starts giving meaningful recommendations, without the need for ten thousand examples. It is the AI ​​that controls the knowledge acquisition process by asking specific questions to salespeople and asking what else it needs. It allows to significantly accelerate the process of acquiring knowledge and obtaining results that coincide with the effect of work of salespeople. Of course, the learning process can be carried out longer, so that the results are even better and get the synergy of combining knowledge from several sellers. 

What does this process look like from the online store customer perspective? The system behaves like a consultant in a stationary store, who before advising, asks the client about the most important aspects necessary to identify the right product. During customer service, feeCOMPASS collects information about its real needs via a simple questionnaire. It does not guess the needs of a particular client by analysing similarities with other clients. Getting to know the customer better is definitely a more effective method of recommending products accurately.

Our AI algorithms, trained by your salespeople, are able to serve personalised suggestions to any customer, even those who visit your store for the first time. Such accurate recommendations help them make the decision. This individual approach affects customer satisfaction, who will gladly return and share their satisfaction with others.

The road does not end here however – it rather opens for new opportunities offered by modern e-commerce. Serving more demanding customers who can be redirected to real advisers. The use of chat – or voice bots that ask about the customer’s needs before he or she even reaches the store’s website. In this way, we smoothly switched to sales processes, which will always benefit from an effective recommendation engine.

If you are interested in feeCOMPASS solution, please learn more about functionalities of our solution and examples of applications in selected online stores.