When an audience is bootstrapped, standard user score computation is not applied, neither is standard vigintile computation.
Only 5 positive possible affinity index values can be computed: 1, 5, 10, 15, 20.
The value will depend on the number of eligible URLs visited by a user on the retroactive bootstrap period, and it depends on a set of integer values that are written in a configuration file - these are arbitrary, but can be tweaked by Data Science / R&D team.
It is therefore guaranteed that a user with an affinity index of 20 on a bootstrapped cluster, will have a higher affinity than a user with an affinity index of 15, 10, etc.
What is absolutely not guaranteed is a homogeneous distribution of audience volumes among boostrapped affinity indexes.
However, if bootstrap + streaming has been ordered, then, after a while (weeks), the volume distribution imputed from streaming will prevail, over the one imputed from bootstrap.
0 Comments