There Is No Single Definition of Customer: Don't Try to Create One

In presentations, articles, and discussions of data governance, semantics - and even Master Data Management - one of the most common examples that is trotted out is the need to have a good definition of “Customer”. It is taken as self-evident that every enterprise needs to have a single definition of Customer that everyone can agree with. 

The problem is that there cannot be a single definition of “Customer”. 

Right away, this seems like it must be wrong. After all, if we are using a term like “Customer”, then there has to be a definition for it, otherwise we would be speaking and writing gibberish. The thesis we are going to defend here is that when people use the term “Customer” there is a clear definition – but only within the context in which they are using the term at that moment. Within any enterprise there are many such contexts, and the same people can even operate in more than one of these contexts. So, there is no single definition of “Customer”.

Some Basic Metaphysics

To better appreciate why there is no such single thing as a “Customer” we need to go back to basic metaphysics.  Aristotle recognized “substances” as things that exist in their own right, and which are the bearers of properties. Although these things exist as individuals (mostly), they are distributed as types in reality. This is what in the data industry we call “entities” or more precisely “entity types”. We don’t use the word “substance” in common speech the way philosophers do, but in the world of data we are pretty familiar with entities.

Now, entity types are referred to using common nouns in English. Data professionals think about nouns too - we sometimes hear that Master Data Management (MDM) is concerned about the nouns of the enterprise. But not all nouns correspond to true entity types (meaning “substances”).   

Think of the ten things you would take out of your house if it caught fire. They might include: the cat; your silver candlesticks; family photographs; and so on. These are all different entity types. What unites them in the “Ten Things You Would Take Out of Your House if It Caught Fire” is your attitude towards them and your desire to manage them a certain way. But this does not make all these different things truly instances of the same type of thing. Rather, they are members of a collection. 

Customer: A Collection of Subtypes

So it is for Customer. “Customer” is really a set of subtypes. For instance, a Customer might be a corporation (a legal entity), or might be an individual human being (a natural person). 

Different subtypes of Customer have different attributes. Or rather, the different subtypes of Customer certainly have many attributes in common, but each subtype has its unique attributes. A Corporation has an Incorporation Date as a unique attribute. An individual human being has a Birth Date as a unique attribute.  

Our job, as data professionals, is to figure out what the subtypes of Customer are, what the common attributes are, and what the unique attributes are. But there is no single type that is a Customer – there is a collection of subtypes, each with their own definition.

Differing Management Scopes

The next fact that we have to consider is that different areas of the enterprise have their own processes, and these processes manage “Customers” in different ways, so that what they think of as a “Customer” varies between them. For instance, Marketing will likely consider prospects to be “Customers”, but for Accounts Receivable, a “Customer” is someone who has paid money to the enterprise, or owes money to the enterprise. 

Thus, different areas of the business decide what is or is not a Customer based on the nature of the work they do. The scope of what each area considers to be a Customer varies so that there is no common definition of Customer. This is a little bit like how everyone has their own unique list of the “Ten Things You Would Take Out of Your House if It Caught Fire”. Again, there will be many different subtypes of Customer as a result of this.

Customer Journeys

Being a Customer involves a process. Often there are phases that form a Customer Journey, such as Suspect, Prospect, Buyer, Alumnus. But even within these phases an individual Customer may interact with different systems or parts of the enterprise in different ways. This results in the Customer’s data being built out differently at different times, and different events happening for the Customer.  This can even happen for what otherwise would be Customers of the same subtype.   The result is even more granular subtypes of Customer. 

Reification and Extrinsic Attributes

Can we really ask what a Customer is without asking what type of thing a Customer is? I would argue we cannot. Here a deeper philosophical problem arises, which finds its practical expression in the Party Model. 

The Party Model is a set of database design patterns that are used for Customer (and other) MDM databases. It points out that, for instance, Corporate Customers can be legal entities, but so can Suppliers, Partners, Brokers, and other kinds of “roles”. Rather than have one data store for each, we should capture the “basic” information for a legal entity in one place – like Legal Name, Jurisdiction of Incorporation, Incorporation Date etc. Then, the attributes that apply to each role (Supplier, Partner, etc.) can be added on in role-specific tables. The same thinking can be applied to Customers that are individual humans.

The metaphysical problem that underpins this is that the real type of thing here is the legal entity. “Customer” is just how an enterprise thinks of a legal entity in the context of a set of relationships and events. It is not a real type of “entity” in the metaphysical sense. The concept of “Customer” can only exist in an enterprise that thinks of something as a “Customer”. This is a case of reification, which means turning something that is not a thing into a thing. 

You and I are Customers of hundreds if not thousands of enterprises, each with its own view of us.  Each of these enterprises has reified us because it makes it easier to manage the information about us that way. But we are not “Customers”, we are individual instances of human beings.

It gets worse. Each of use has intrinsic attributes. These are attributes that we possess irrespective of what any mind thinks about us. E.g., height, mass, birth date. But Customers have a lot of extrinsic attributes as well. These are attributes that created by enterprises and are not intrinsic. I probably have a Customer Lifetime Value for Amazon, Microsoft, Target, and a myriad of other enterprises – but none of these attributes are inherent to me.  There is no space to go further into this, but the modern logician P. T. Geach explored this area and should be read for an introduction.

Conclusion

You cannot define something that does not exist in a real sense by using traditional approaches to definitions. To exist in a “real” sense means that something must exist independently of minds and be an Aristotelian substance.  “Customer” does not exist in this way and is not an Aristotelian substance. Rather, Customer is a mind-based construct.  It still exists, but not in the way we usually think things exist. Because it is mind-dependent, different parts of the enterprise will genuinely see Customer in different ways. They will each have their own definition, and these definitions are true for them. This is not postmodernism where feelings and opinions are “truth”. Each viewpoint of what a Customer is is genuinely based on how that part of the enterprise interacts with Customers. 

So there is no single definition of Customer. One wonders why people could ever think there should be.