Systems and methods for intelligent promotion design with promotion selection

Abstract

Systems and methods for the selection of the promotions are provided. A set of possible offers are initially received, each including a set of variables, with each variable having a set of possible values. These form a combination of variable values for each offer. A heuristic is applied to all possible offers to reduce the number of offers being considered. The combination of variable values for these reduced number offers is converted into a vector value, which is then scored, ranked and the top ranked offers are selected for inclusion in a promotional campaign. The remaining offers are then analyzed to select additional offers to include into the promotional campaign which maximizing a determinant for the selected offers using their vectors. All selected offers are administer in a promotional test campaign across many consumer segments. Feedback from the campaign may be collected to generate a “general” offer.

What is claimed is: 
     
         1 ) A method for selecting promotions for a test campaign comprising:
 receiving a plurality of possible offers comprising a plurality of variables, each variable having one of a set of values to form a combination of variable values;   applying heuristics to the plurality of possible offers to generate a reduced set of possible offers;   converting the combination of variable values for each of the offers into a vector;   scoring the reduced set of offers based upon the vector of each offer;   ranking the reduced set of offers by their scores;   selecting a first subset of the ranked reduced set of offers based upon the offers with the highest scores;   selecting a second subset of the ranked reduced set of offers, wherein the offers in the second subset are mutually exclusive to the first subset, and further wherein the second subset is selected by maximizing a determinant for the subset using the vector for the offers in the second subset; and   administering a promotional test campaign using the first and second subset of offers.   
     
     
         2 ) The method of  claim 1 , further comprising collecting feedback from the administered promotional test campaign. 
     
     
         3 ) The method of  claim 2 , wherein the administering is tested across a plurality of consumer segments. 
     
     
         4 ) The method of  claim 2 , further comprising generating a general promotion campaign based upon the collected feedback. 
     
     
         5 ) The method of  claim 1 , wherein the heuristics include rounding rules. 
     
     
         6 ) The method of  claim 1 , wherein the heuristics include predefined rules corresponding to goals, and further wherein a user selects one goal and the associated predefined rules are applied to the set of offers. 
     
     
         7 ) The method of  claim 1 , wherein the scoring includes:
 receiving a set of training offers comprising a plurality of variables, each variable having one of a set of values to form a combination of variable values;   converting the combination of variable values for each of the training offers into a vector;   generating pairings of the training offers such that all combinations of training offer pairs is represented;   subtracting the vector of one training offer in each pair from the other vector of the other training offer of the pair to generate a pair vector;   collecting success metrics for each of the training offers from a retailer's point of sales system, a computerized application, and from consumer mobile devices;   subtracting the success metrics of the one training offer in each pair from the other success metrics of the other training offer of the pair to generate a raw score;   generating a normalized score for each of the pairings using the raw score and the pair vector;   generating a model, by machine learning, using the normalized score, wherein the model is one of a decision tree and a neural network; and   applying the model to the set of reduces offers using linear regression to generate the scores.   
     
     
         8 ) The method of  claim 1 , wherein the determinate is calculated by generating a matrix X of the vectors of the offers being considered for the second subset. 
     
     
         9 ) The method of  claim 8 , wherein the determinate is calculated for (X T X). 
     
     
         10 ) The method of  claim 1 , wherein the size of the first subset and second subset are equal. 
     
     
         11 ) The method of  claim 1 , wherein the first subset consists of 3-10 offers. 
     
     
         12 ) The method of  claim 1 , wherein the second subset is consists of 3-15 offers. 
     
     
         13 ) The method of  claim 1 , wherein the second subset is selected from the top 20-30 ranked offers after the first subset of offers.