Method and apparatus for determining promotion pricing parameters

Abstract

A method, apparatus, and computer program product are disclosed to improve selection of promotion pricing parameters. The method may determine one or more promotion pricing parameters for a promotion that is offered by a promotion and marketing service. The method includes generating one or more predictive models based on historical promotion performance data and generating a revenue equation using the one or more predictive models. The revenue equation provides an estimate of a revenue received by the promotion and marketing service based on the one or more predictive models. The method further includes determining an estimated revenue using the revenue equation based on one or more input sets of promotion pricing parameters provided as input to the revenue equation, and selecting at least one of the input sets of promotion pricing parameters for the promotion based on the estimated revenue. A corresponding apparatus and computer program product are also provided.

1 .- 51 . (canceled) 
     
     
         52 . An apparatus for determining one or more promotion pricing parameters for a promotion, the apparatus comprising a processor and a memory including computer program code, the memory and the computer program code configured to, with the processor, cause the apparatus to:
 generate, via a promotion performance model generation module, two or more predictive promotion performance models based on historical promotion performance data, wherein each predictive promotion performance model is generated as a result of a machine learning algorithm that uses the historical promotion performance data as a model training set;   generate, via a revenue equation generation module, a revenue equation based on at least one of the generated predictive promotion performance models;   determine, via a promotion pricing module, one or more promotion pricing parameter sets using the revenue equation and a merchant promotion input value received from a merchant device, wherein at least one promotion pricing parameters set comprises a modified promotion pricing parameter and an additional promotion pricing parameter, the modified promotion pricing parameter and the additional promotion pricing parameter derived from the merchant promotion input value based on the revenue equation;   identify a selected promotion pricing parameters set;   generate, by the processor, an active promotion based on the selected promotion pricing parameters set;   monitor, by the processor, a performance characteristic of the active promotion; and   update at least one of the predictive promotion performance models based on the performance characteristic of the active promotion, such that the predictive promotion performance models and the revenue equation are continually refined via a positive feedback loop to provide improved predictions of optimal promotion pricing parameters.   
     
     
         53 . The apparatus of  claim 52 , wherein the revenue equation is generated for a particular merchant associated with the merchant device. 
     
     
         54 . The apparatus of  claim 52 , wherein the revenue equation is generated for a particular type of merchant associated with the merchant device. 
     
     
         55 . The apparatus of  claim 52 , wherein the revenue equation is generated for a particular promotion category. 
     
     
         56 . The apparatus of  claim 52 , wherein identifying the selected promotion pricing parameters set comprises:
 causing presentation of the one or more promotion pricing parameter sets to the merchant device; and   receiving an approval notification from the merchant device, the approval notification associated with the selected promotion pricing parameters set.   
     
     
         57 . The apparatus of  claim 52 , wherein at least one of the one or more promotion pricing parameter sets is optimized for bringing in new customers. 
     
     
         58 . The apparatus of  claim 52 , wherein at least one of the one or more promotion pricing parameter sets is optimized for maximizing overspending beyond the promotional value of the active promotion. 
     
     
         59 . The apparatus of  claim 52 , wherein at least one of the one or more promotion pricing parameter sets optimizes merchant profit based on a promotion redemption value. 
     
     
         60 . The apparatus of  claim 52 , wherein the modified promotion pricing parameter is derived from the merchant promotion input value to maximize the revenue equation. 
     
     
         61 . The apparatus of  claim 52 , wherein the additional promotion pricing parameter is derived from the merchant promotion input value as a promotion pricing parameter not received from the merchant device. 
     
     
         62 . The apparatus of  claim 52 , wherein the historical promotion performance data comprises, for each of one or more past promotions, a promotion type, a merchant category, a discount level, an accepted promotion value, a promotion date range, a number of received promotion impressions, a number of offered promotions, a promotion redemption rate, and a promotion refund rate. 
     
     
         63 . The apparatus of  claim 52 , wherein the performance characteristic comprises at least one of a promotion redemption rate, a promotion size, or a promotion refund rate. 
     
     
         64 . The apparatus of  claim 52 , wherein at least one of the two or more predictive promotion performance models is a demand model, the demand model configured to predict an expected demand for the promotion, wherein the demand model is trained using the historical promotion performance data as a demand training set, such that training includes predictively analyzing an impact of various promotion parameters on a promotion size of past promotion offerings. 
     
     
         65 . The apparatus of  claim 64 , wherein the demand model includes a regression analysis of the historical promotion performance data. 
     
     
         66 . The apparatus of  claim 65 , wherein the regression analysis is calculated in accordance with an equation to predict a promotion size:
   log  q=α   1  log  p+α   2  log  d+α   3   c+α   4   sc+α   5   ds+α 6 di+α   7   r+α   8      and each of the α values are constants weights to be derived via the regression analysis, p is a unit price, d is a discount, c is a merchant category or type, sc is a merchant subcategory, ds is a promotion service category, di is a division, and r is a merchant quality score.   
     
     
         67 . The apparatus of  claim 66 , wherein for a given merchant, each of the merchant category or type, the merchant subcategory, the division, the promotion service category, and the merchant quality score is a known and fixed factor such that particular portions of the equation to predict the promotion size are constant, and after accounting for the known and fixed factors, the promotion size of a particular promotion is calculated by:
   log  q=α   1  log  p+α   2  log  d+α   0      and wherein α 0  is a constant representing the known and fixed factors for the particular promotion.   
     
     
         68 . The apparatus of  claim 52 , wherein at least one of the two or more predictive promotion performance models is a margin model to determine an expected margin for a promotion and marketing service for each potential sale of the active promotion, wherein the margin model is trained using the historical promotion performance data as a margin training set, such that training includes predictively analyzing an impact of various promotion parameters on merchant return-on-investments (ROIs) of past promotions and predicting a maximum margin available to a promotion and marketing service to ensure a threshold ROI for a merchant associated with the merchant device. 
     
     
         69 . The apparatus of  claim 68 , wherein the expected margin is a portion of an accepted value received by the promotion and marketing service, that ensures (i) at least the threshold ROI for the merchant, such that when the proposed promotion is redeemed by a consumer towards a purchase of particular goods, services or experiences offered by the merchant, the merchant receives the threshold ROI from the promotion and marketing service to account for at least a portion of a price of the particular goods, services or experiences provided to the consumer, while concurrently (ii) establishing that a minimum amount of revenue is generated by each sale of the active promotion. 
     
     
         70 . The apparatus of  claim 68 , wherein the margin model includes a regression analysis of the historical promotion performance data. 
     
     
         71 . The apparatus of  claim 70 , wherein the regression analysis is calculated in accordance with an equation to predict a threshold margin:
     b*=β   1   p+β   2   d+β   3   c+β   4 cog+β e  
   and each of the β values are coefficients to be derived via the regression analysis, b* is the threshold margin that makes a merchant return-on-investment (ROI) greater than or equal to a threshold value, p is a unit price, d is a discount, c is a merchant category or type, and cog is a percentage of a promotion price of particular goods, services or experiences provided to a consumer in exchange for the active promotion.   
     
     
         72 . The apparatus of  claim 71 , wherein for a given merchant, one or more factors of the proposed promotion is a known and fixed factor such that particular portions of the equation to predict the threshold margin are constant, and after accounting for the known and fixed factors, a threshold margin size of a particular promotion is calculated by:
     b*=β   1   p+β   2   d+β   0      and wherein β 0  is a constant representing the known and fixed factors for the particular promotion.   
     
     
         73 . A computer-implemented method comprising:
 generating, via a promotion performance model generation module, two or more predictive promotion performance models based on historical promotion performance data, wherein each predictive promotion performance model is generated as a result of a machine learning algorithm that uses the historical promotion performance data as a model training set;   generating, via a revenue equation generation module, a revenue equation based on at least one of the generated predictive promotion performance models;   determining, via a promotion pricing module, one or more promotion pricing parameter sets using the revenue equation and a merchant promotion input value received from a merchant device, wherein at least one promotion pricing parameters set comprises a modified promotion pricing parameter and an additional promotion pricing parameter, the modified promotion pricing parameter and the additional promotion pricing parameter derived from the merchant promotion input value based on the revenue equation;   identifying a selected promotion pricing parameters set;   generating an active promotion based on the selected promotion pricing parameters set;   monitoring a performance characteristic of the active promotion; and   updating at least one of the predictive promotion performance models based on the performance characteristic of the active promotion, such that the predictive promotion performance models and the revenue equation are continually refined via a positive feedback loop to provide improved predictions of optimal promotion pricing parameters.