July 24, 2017

Circulation Theory, Part 2



I asked you to predict what you might see if you mailed Segment 8, Segment 9, and Segment 10.

What was your guess?

I like to "cumm" results and then fit a line to cumm'd results. The relationship looks something like this.


With cumm'd results, I then subtract the difference ... the prediction for Segment 8 in the graph minus the prediction for Segment 7 yields the dollar per book for Segment 8.

In fact, I'll do that for every data point - which allows me to smooth out the problems with noisy actual results (the relationship between Segments 3/4/5 isn't quite perfect due to noise, for instance).

This leaves me with the following table.


Just like that, I have a reasonable prediction for each segment, with the noise smoothed out!

Ok - here's your quiz question for tomorrow ... if you executed a test of a 64 page catalog against the 128 page catalog, and you learned that 64 pages generated 75% of the demand of a 128 page catalog, how would you fill in each of the remaining cells in the grid (for 32 pages, 64 pages, and 96 pages)?

Do your homework, and we'll review results tomorrow, ok?