Why not use survey data instead of my client data? I am often asked this question ... here is a short explanation ...
When we talk about customer data, we refer to transactional data, the richest piece of information, but unfortunately in the majority of cases little or badly used by companies.
So, why? Much less complicated to use a representative survey result from my client base than to extract and process data from I do not know where ... billing history, customer purchase transactions, SKU, dates and purchase site location, payment method, etc. ; in fact, by customer data, we are talking about all the information associated with your client's relationship with you, the company ...
Initially, the survey data is by definition static and not dynamic; a photo taken at a specific moment in time. In general, the information entered is grouped into dichotomous (yes, no) and ordinal (many, moderately, a little less or not at all ...). In contrast, the transactional data is mainly of a chronological and continuous nature ... "a metric interval is required, among other things, for Euclidean distance analyzes" Survey data comes, by definition, with a margin of error associated with the respondent's limited memory and the objectivity of its answers. What were you doing yesterday, OK, good memory, the day before yesterday, and three months ago 2 days and 15 minutes? Hey, are you saying how much beer did you drank that night? In the case of the transactional data, it has the infinite memory of the cash register and no problem of objectivity ...
The majority of data modeling analyzes, exploratory and confirmatory, parametric and non-parametric analyzes can be done with transactional data, but it is much more limited for survey data ... In some cases, we can combine survey data with client transactional data ... and we can then perform some exploratory analyzes, Khi2 association tests ... analysis of variance, correlation, decision tree, etc. But let's say, without a shadow of a doubt, the power of transactional data, combined with enrichment data is really the great wealth of information that should be taken advantage of.
An earlier version of this text was published on the blog Analytical Marketing by Guy Mercier WordPress on August 23, 2011.