The process of categorizing records that will be mailed or telemarketed into test cells is called segmentation. The segmentation process splits the records randomly or based on other attributes, such as demographic or firmographic data, into cells, and assigns each cell a keycode. If the keycode can be captured during the response, then the number of responses can be measured against the quantity mailed by keycode and a response rate can be determined.

A typical segmentation matrix might look like:

Lists:\Packages:	AB	DN	KC	PO	Total
X1	21,081	21,081	21,081	21,081	84,324
X2	1,776	1,776	1,776	1,776	7,104
X3	1,426	1,426	1,426	1,426	5,704
X4	3,225	3,225	3,225	3,225	12,900
Y1	11,686	11,686	11,686	11,686	46,744
Y2	2,687	2,687	2,687	2,687	10,748
Y3	2,424	2,424	2,424	2,424	6,696
Y4	12,695	12,695	12,695	12,695	62,780
Total	60,000	60,000	60,000	60,000	240,000

Notice that in this particular segmentation, equal amounts of records were desired for each test package (each column has the same number of records).