An Overview of Data Mining Techniques - Data Mining Techniques in CRM: Inside Customer Segmentation - page 31

Figure 2.6 Lift chart.

The Maximum Benefit Point

An approach often referred to in the literature as a rule of thumb for selecting
the optimal size of a targeted marketing campaign list is to examine the gains
chart and select all top tiles up to the point where the distance between
the gains curve and the diagonal reference line becomes a maximum. This
is referred to as the maximum benefit point and it is the point where the
difference between the gains curve and the diagonal reference line has its
maximum value. The reasoning behind this approach is that, from that point
on, the model classifies worse than randomness. This approach usually yields
large targeting lists. In practice analysts and marketers should take into
consideration the particular business situation, objectives, and resources and
possibly consider as a classification threshold the point of lift maximization.
If possible, they should also incorporate in the gains chart cost (per offer)
and revenue (per acceptance) information and select the cut-point that
best serves their specific business needs and maximizes the expected ROI
and profit.