KNOWING BEFORE IT HAPPENS_ Artificial Intelligence makes retailing more efficient
Traditionally, it is understood that a segment is a part of the market made up of people sharing a similar profile, which means they can behave in a similar way if they are exposed to some marketing policies.Nevertheless, thanks to the evidence provided by Artificial Intelligence (AI) applied to retailing it is being proved that the same individual can simultaneously belong to several segments, but with a different degree of adequacy to each one.
Today, AI programs allow us to combine qualitative and quantitative data, which in turn enables us to get much closer to the complexity of human life and therefore to predict short-term customer behaviour.
Do you spend as much money on holidays as you do on any other working day? Are you equally sensitive to prices in both situations?
Chances are that you behave differently depending on each situation. However, even though your behaviour is at odds with the conventional theory of segmentation, you don’t need to go to the shrink . That happens to almost everyone.
Reviewing the idea of segmentation
Traditionally, it is understood that a segment is a part of the market made up of people sharing a similar profile, which means that they can behave in a similar way if they are exposed to specific marketing policies. Therefore, each segment must be different from the others.
Volumes have been written stating that for segmenting, a number of criteria can be applied such as socio-demographic (age, gender, spending power, etc. which are so often cited, yet actually of little use ), psychographic (including attitudes, values, etc.), and even behavioural (for instance shopping habits, …).
Yet, if you in fact behave differently depending on the context or situation you find yourself in, you are actually saying that the traditional theory has its shortfalls. And I believe you are right.
Today there is the possibility of using Artificial Intelligence (AI) programs applied to retail is validating the intuition that the reality differs from the conventional theory, as my ESADE colleague Monica Casabayo, explains. The same individual can simultaneously belong to several segments, but having different degrees of adequacy to each one of those segments.
If we apply this, for example, to the beauty centres retail industry, customers can be perfectly segmented by anticipating the level of their economic value as customers, based on what was their first contact with the store. For instance, it could have been a discounted hair removal service, a recommendation from a friend, etc.
One of the main advantages of this method is that we can simultaneously work with quantitative (volume of purchase, number of products, etc.) and qualitative (complaints, type of promotion, preferences for brands, etc.) data. This enables us to get much closer to the complexity of human life.
AI can also be applied to retail to find out how many employees would be required at the checkouts next Saturday, taking into account the weather forecast and the fact that there is a football game scheduled between the two major local clubs.
In short, by using AI programs, retail companies achieve something that statistic applications can only dream of: ex ante results. That is, results that can predict what is going to happen in the near future with a high degree of probability. Statistics can only explain what has already taken place.
Another example of the use of AI has to do with its ability to detect customers with disloyalty risk, when it is already known that a new competitor is arriving soon to the store’s catchment area. Once the list of those customers is completed, the company’s next step would be to devise and apply a counteracting loyalty reinforcement plan.
The two main requirements for using an AI applications are having access to professional specialists and to a large quantity of properly gathered data from past periods.In that sense retail companies are magnificent: the amount of raw data they generate is huge.
People (and segments) do change
A further advantage offered by this AI tool is that it provides solutions to a major issue: customer databases become obsolete in a blink. When the eldest daughter in a household, holder of a customer card, moves in with her boyfriend, her mother won’t come to the store to let us know that we have to change the number of people living under that roof so that we can recalculate better our market share per household.
Using AI, the company is based on facts and real behaviour, and it builds on that to understand its customers.
Don’t trust what your customer says he does, and even less about what he says he is going to do. Actions speak louder than words.
Source: Distribución Actualidad, the spanish retail magazine
(nº 415, may 2010)