In a recent article [Testing – the most effective tool for database marketing], we have analysed the importance of testing in marketing campaign management.
Testing is done by promoting an offer to a smaller group of customers in order to evaluate the effectiveness of the whole approach (value proposition, message used, timing, target group). This smaller group is a sample which is used to predict the behavior of a larger group. Sampling always introduces bias in any measurement. The larger the sample is, the smaller the bias introduced.
Assuming that: • an expected positive response rate p is 2 % (the percent of customers accepting the offer - this estimate may be an average of previous similar tests), • an expected standard deviation S 0,05 % in response rates. This estimate may be derived from previous tests: S can be calculated after executing a campaign on a group of given size (actually, the accurate term for this is ‘standard error of a proportion’ in a binomial distribution which however resembles highly the normal distribution),
a very simple formula to estimate the size of the needed test group is the following: Group size = p * (1-p) / S2 , and yields a value of 78.400.
Should a sensitivity analysis be carried out by changing the input values, the results do not provide a feeling of a stable estimation. For example if the expected positive response rate p is 1,5 % and the expected standard deviation S 0,06 % then the group size should be 41.042. The sensitivity analysis helps recognise the importance of the estimations used. A relatively small change leads to a huge difference.
Therefore, Campaign managers should analyse the findings of previous campaigns, in order to have a good idea on the expected range of p and S values. Previous findings contribute to the adjustment of future test group sizes. Though this may not be a highly safe approach in dynamically changing market conditions and offers, it is much better than using randomly selected group sizes, which may lead either to inaccurate findings or wasted resources when the group size is much higher than needed.
Concluding, in order to build campaign management intelligence, a Business should capture and analyze the results of each campaign or test campaign. The evaluation of these results shall assist in the optimization of future campaigns.
Home page - Customer Intelligence