Facts Anchors

Patterns of Analysis: Facts “Anchors”

Numbers by themselves mean little, so is easy to do a data pull and summarize, without almost noticing and realizing what each number means, this is of course not good, limits the potential of a good insight and also lets more errors creep in.

An approach to help with that, is to, even at the data pull phase of the analysis, start noticing facts and making them anchors to rest of analysis, like total users is about 10k, the ones on Sunday are about 1k, weekend users behave very differently than the ones on the workdays, notice these anchor references and keep them in mind, because for every new number pulled, these references can be used both as a validation of new pulled data (looking at different angles to catch errors) and as a leverage to build upon for deeper insights.

Also, if there is any other reference available, like an existing report, or a comparable know metric, etc go look at it, make sure your data will match close enough that one for that time frame. - this is a great data validation technique.

Modeling

This is really just like modeling, we are here forming a model based on observed facts about the type of user, or type of behavior, or type of product performance, etc…