Ian, that is an excellent debate to open! Arca Blanca works with a variety of retailers across Europe, and this is a very live discussion in many quarters. To add to your points the answers will very much depend on the question. While this may be a truism, consider it in the context of i) what makes a good and valuable customer? ii) what is it that the customer values? and iii) what are their characteristics?
When those questions are asked there is often an assumption about the answer, such as the financial life-time-value of a customer or what the demographics look like – in effect limited by what I would put into said spreadsheet in the first instance. However, increasingly we use data science to help retailers identify much more complex patterns, in combinations that we would be unlikely to imagine ourselves. Like the company which found that customers are 7 times more valuable if they live in a location with certain characteristics, shop in a town with a combination of other characteristics (competition, parking, etc), are of a certain demographic and have access to a certain format with a specific range and service level. The number of combinations is almost limitless and this is where AI will offer retailers access to a whole new set of insights – not necessarily to make decisions for them (that is an entirely different – and equally interesting discussion), but to help them evaluate what matters for their brand and hence where to invest.
Thanks Gitte, a really interesting point. I did find myself wondering whether a more unstructured data science approach might be a way of answering (for example) the question about counters in Tesco. That still leaves the big questions about what you measure and therefore what goes into the model, but it is a potential way of exposing relationships that might not otherwise seem obvious.
Ian, that is an excellent debate to open! Arca Blanca works with a variety of retailers across Europe, and this is a very live discussion in many quarters. To add to your points the answers will very much depend on the question. While this may be a truism, consider it in the context of i) what makes a good and valuable customer? ii) what is it that the customer values? and iii) what are their characteristics?
When those questions are asked there is often an assumption about the answer, such as the financial life-time-value of a customer or what the demographics look like – in effect limited by what I would put into said spreadsheet in the first instance. However, increasingly we use data science to help retailers identify much more complex patterns, in combinations that we would be unlikely to imagine ourselves. Like the company which found that customers are 7 times more valuable if they live in a location with certain characteristics, shop in a town with a combination of other characteristics (competition, parking, etc), are of a certain demographic and have access to a certain format with a specific range and service level. The number of combinations is almost limitless and this is where AI will offer retailers access to a whole new set of insights – not necessarily to make decisions for them (that is an entirely different – and equally interesting discussion), but to help them evaluate what matters for their brand and hence where to invest.
Thanks Gitte, a really interesting point. I did find myself wondering whether a more unstructured data science approach might be a way of answering (for example) the question about counters in Tesco. That still leaves the big questions about what you measure and therefore what goes into the model, but it is a potential way of exposing relationships that might not otherwise seem obvious.