…will terminate a subscription or not, a customer will purchase $X, etc.  Lift is a measure of how much better the statistical model does than not using a model at all.  Decile lift is this measure applied to deciles of the target records ranked by predicted probability (for a binary outcome) or predicted amount (for a continuous variable).  For the top decile lift the steps, for a 0/1 classification problem, are

1.  Split records into training and validation samples

2.  Train a model on the training data, apply it to the validation data

3.  Rank the validation data in order of predicted probability of being a “1”

4.  Count the number of actual 1’s in the top decile of the validation data

5.  The lift is the ratio of #4 to the average number of 1’s per decile across the entire validation set