There are a lot of really smart people out there who do attribution work. They use complex statistical models to try and tie orders back to the advertising that drove the order.
But attribution models have one fatal flaw ... they incorrectly attribute orders that would happen organically (i.e. without advertising) back to advertising activities, causing us to spend way too much money on marketing activities.
If you agree with me that we're over-attributing orders to marketing activities ... then give this presentation a read, where I'll share with you a mail/holdout methodology for properly assigning attribution rules.
Subscribe to:
Post Comments (Atom)
Seasonality is Amplified
Back to our table from yesterday, reviewing customers with two life-to-date purchases, measuring how they perform in the next month based on...
-
It is time to find a few smart individuals in the world of e-mail analytics and data mining! And honestly, what follows is a dataset that y...
-
It's the story of 2015 among catalogers. "Our housefile performance is reasonable, but our co-op customer acquisition efforts ar...
-
This is where we're headed: Let's say you want to invest an additional $100,000 in paid search. You should be able to see a p&l,...
How did you come up with the attribution rules (for example the one about "Attribute $1.00 of every $4.75 online to the catalog (21% of online dollars)"? From the hold out test?
ReplyDeleteYou bet, that's exactly where the results came from! And when they come from a controlled experiment, then you can trust them a whole lot more.
ReplyDeleteKevin - From the numbers shown?
ReplyDeletePages 11, 12, and 17 have all of the numbers that source the comments in the presentation.
ReplyDeleteThanks,
Kevin