One specific comment, others that echo this, and ~my own experience:
Someone with experience in email marketing found “most of the insights from data-driven analysis to be unreliable.” … many reports on ‘best time of day to send’ contradict one another.
A large multi-dimensional space of choices to explore, many combinations to test, hard to know how to generalize.
Tracking and attribution issues
Ad-optimization algorithm (e.g, Facebook) are not transparent and can be confusing to draw inferences from
I think it’s partly a reaction to for-profit marketers overselling themselves, and partly a statement of ’this being a hard task in general.
Wonderful write-up, David. Thank you for this!
Could you expand on what you think is meant by: Unreliability of existing work in ‘data-driven marketing’?
One specific comment, others that echo this, and ~my own experience:
Someone with experience in email marketing found “most of the insights from data-driven analysis to be unreliable.” … many reports on ‘best time of day to send’ contradict one another.
A large multi-dimensional space of choices to explore, many combinations to test, hard to know how to generalize.
Tracking and attribution issues
Ad-optimization algorithm (e.g, Facebook) are not transparent and can be confusing to draw inferences from
I think it’s partly a reaction to for-profit marketers overselling themselves, and partly a statement of ’this being a hard task in general.