Linkage to reduce errors in online promotion testing
Methods and apparatus for improving the result of online promotion testing are disclosed. In one or more embodiments, base-lining is employed to improve the mapping between the online test promotion responses and projected in-store result. Alternatively or additionally, the test population responses may be adjusted to account for differences between the test population and the in store customer population.
1. A computer-implemented method for relating a variable values to elasticity curves comprising:
generating a plurality of tests for online platforms, wherein the plurality of tests are generated for permutations of variable values;
administering the plurality of tests to a plurality of segmented subpopulations of individuals on the online platforms;
obtaining responses from said segmented subpopulations of individuals, wherein the responses include click-through rates;
generating a best fit curve between the click-through rates and the variable values;
extrapolating a plurality of data points from the best fit curve;
linking the best fit curve to a known elasticity curve using statistical regression using the extrapolated plurality of data points to generate a predictive model of volume lift in a store versus click-through rates, wherein the elasticity curve is a measure of in-store lift versus the variable value; and
predicting actual in-store lift based upon click-through rates for a a set of variable values using the predictive model.
2. The method of claim 1 , further comprising generating the general promotion responsive to the obtained response.
3. The method of claim 2 , wherein the general promotion is generated responsive to the linkage model for a desired lift.
4. The method of claim 1 , wherein the linking of the best fit curve to the elasticity curve includes statistical regression.
5. The method of claim 1 , wherein the elasticity curve is derived from archive data of past promotions for in stores.
6. The method of claim 1 , further comprising stratifying segmented subpopulations of individuals and in-store consumer populations by at least one profile criteria.
7. The method of claim 6 , further comprising adjusting the linking responsive to the stratified segmented subpopulations of individuals and in-store consumer populations.
8. The method of claim 7 , wherein the adjusting comprises:
stratifying the responses based upon the stratification of the segmented subpopulations of individuals;
weighting the stratified responses to reduce error between each response strata and predicted responses of the elasticity curve; and
summing the weighted stratified responses to generate an adjusted click-through rate.
9. The method of claim 8 , wherein the best fit curve is between the adjusted click-through rate and the variable values.
10. The method of claim 8 , further comprising calculating a standard error for the adjusted click-through rate to generate a confidence interval.