pricing analytics
TRANSCRIPT
INFORMATION FOR MANAGEMENT DECISIONS
PRICING ANALYTICS
Pricing Decision Process
Problem
Question
Information Decision
From Information to Decision
Analytical Approach 3 Levels
Analytical Approaches
Level I Pricing Effects Level II Promotion Level III Predictive Analytics
Pricing Effects Introduction Level
Sales Trend vs. Competition Breakeven Analysis Price Elasticity Price-Mix-Analysis
Sales Trend vs. Competition Problem/Issue
Sales last month down 20% YOY while Price 15% higher than all competitors
Does higher price cause the sales decline?
Pricing Effects
Sales Trend vs. Competition Analysis
Our Price Comp A Comp B Sales YOY % Change
Sales Declining
Pricing Position Not Change
Pricing Effects
Sales Trend vs. Competition Lesson Learned
No correlation between sales decline and higher pricing position Always analyze sales and pricing trends. Sales or pricing snapshot often leads to a wrong conclusion.
Pricing Effects
Breakeven Analysis Problem/Issue
5% of price reduction increase sales by
30% Is it a price change successful?
Pricing Effects
Breakeven Analysis Analysis
Not necessary! Current GP = 10% Price Reduction = 5% GP Afterward = 5% Need to double the sales to breakeven!
Pricing Effects
Breakeven Analysis Lesson Learned
Breakeven analysis should base on margin not sales. Price Elasticity affects sales not margin
Pricing Effects
Price Elasticity Problem/Issue
How to determine incremental sales prior to price drop?
Price Elasticity or Pricing Sensitivity
Analysis
Pricing Effects
Price Elasticity Analysis
Price Quantity $ 5.00 881 $ 4.80 956 $ 5.00 881 $ 5.50 728 $ 5.00 881 $ 4.70 997 $ 5.40 755 $ 4.90 917 $ 5.30 784 $ 5.40 755 $ 5.30 784 $ 4.80 956 $ 4.70 997 $ 5.30 784 $ 4.70 997 $ 4.90 917 $ 5.20 815 $ 5.30 784 $ 5.30 784 $ 5.30 784 $ 5.00 881 $ 5.00 881 $ 5.10 847 $ 5.00 881 $ 4.70 997 $ 4.60 1041
Ln(P) Ln(Q) $ 1.61 6.78 $ 1.57 6.86 $ 1.61 6.78 $ 1.70 6.59 $ 1.61 6.78 $ 1.55 6.90 $ 1.69 6.63 $ 1.59 6.82 $ 1.67 6.66 $ 1.69 6.63 $ 1.67 6.66 $ 1.57 6.86 $ 1.55 6.90 $ 1.67 6.66 $ 1.55 6.90 $ 1.59 6.82 $ 1.65 6.70 $ 1.67 6.66 $ 1.67 6.66 $ 1.67 6.66 $ 1.61 6.78 $ 1.61 6.78 $ 1.63 6.74 $ 1.61 6.78 $ 1.55 6.90 $ 1.53 6.95
Coefficients Standard Error t Stat P-value Intercept 10.00 0.00127 7859.64492 0.00002
1.609437912 -2.00 0.00079 -2544.94319 0.00012
ln(Q) = 10 - 2 ln(P) Price Elasticity: -2
Pricing Effects
Price Elasticity Lesson Learned
Sell Price vs. Quantity Data
SAS, R or other tools for more products Log-Linear Demand /Price Model Make sure PE is statistically significant
Pricing Effects
Pricing Mix Problem/Issue
Average Selling Price dropped in the last three months!
But there is no pricing action What happened?
Pricing Effects
Pricing Mix Analysis
Aggregated level price changes can happen due to product mix or channel mix changes.
Sales Units Price 2014 2015 2014 2015 2014 2015 Product A $ 50,000 $ 40,000 500 400 $ 100 $ 100 Product B $ 20,000 $ 30,000 1000 1500 $ 20 $ 20 Total $ 70,000 $ 70,000 1500 1900 $ 47 $ 37
Pricing Effects
Pricing Mix Lesson Learned
Price change could result from the multiple mix changes such as product mix, channel mix and segment mix It is very common!
Pricing Effects
Promotion Evaluation Medium Level
Potential Incremental Sales Baseline Forecast Margin Exposure Breakeven Analysis Test/Control Group Setup
Potential Incremental Sales Problem/Issue
Predict Sales Increase Eliminate Unreasonable Expectations
Promotion Evaluation
Potential Incremental Sales Analysis
Price Elasticity Competition Market Share Previous Promotion Wishful Thinking
Promotion Evaluation
Potential Incremental Sales Lesson Learned
Difficult but not impossible! Predict sales range and possibility Try to eliminate unreasonable thinking
Promotion Evaluation
Baseline Forecast Problem/Issue
Forecast without promotion Trends prior to promotion Seasonality Current market share
Promotion Evaluation
Baseline Forecast Analysis
Excel based forecast – analytic tools Time series forecasting – more than 10 statistical models Multivariable regression models
Promotion Evaluation
Baseline Forecast Lesson Learned
Process commonly ignored Critical step for promotion revaluation Make sure your promotion sponsor agrees Baseline Forecast prior to the promotion
Promotion Evaluation
Margin Exposure Problem/Issue
Price Drop = Profit Loss or Margin Leakage Margin Exposure accesses the potential risk
Promotion Evaluation
Margin Exposure Analysis
Identify the exposing areas Assume worst possible outcomes Calculation of margin loss needs to be done at sales order or invoice level not at aggregate level
Promotion Evaluation
Margin Exposure Lesson Learned
Always ask a following question: Can we afford to lose that much money if the promotion goes south? Average selling price should never be used in calculating margin exposure
Promotion Evaluation
Control Group Problem/Issue
Have the consensus on promotion evaluation and eliminate future dispute
Be prepared for unexpected changes of
market place
Promotion Evaluation
Control Group Analysis
Blinded Group: Similar set of promotion targets Preselected Group: The baseline group selected by promotion sponsor Randomized Group: Randomly divide the set of targets into test and control groups ANOVA Test: Statistical significant or not?
Promotion Evaluation
Control Group Lesson Learned
Less controversy and commonly adopted by promotion sponsor and pricing professional Best promotion evaluation approach!
Promotion Evaluation
Predictive Analytics Advanced Level
Outlier Identification Association - Shopping Basket Analysis Principal Component Analysis Margin Bridge Analysis Classification – Decision Tree
Outlier Identification Problem/Issue
Outliers often mislead analysis and promotion evaluation
Always try to eliminate them prior to any
analytical approach
Predictive Analytics
Outlier Identification Analysis
Predictive Analytics
IQR Rule: Q3+1.5x(Q3-Q1) Regression Model (Robust, Cook-D etc.) SAS, SPSS, R and SQL Data Mining Tool
Outlier Identification Lesson Learned
Outliers are segment/cluster driven Single variable based outliers vs. multi-variable based outliers The outlier in one segment might not be the outlier in another segment
Predictive Analytics
Association Problem/Issue
Opportunity for effective cross-sell Critical step to price product bundles Understand price elasticity between anchor
and attach Discount applied to targeting anchor not
attach
Predictive Analytics
Association Analysis
Predictive Analytics
Excel Data Mining Tool Add In SAS Enterprise Miner/R
Association Lesson Learned
Multiple levels of associations could occur Could well miss your shot i f anchor and attach are f l ipped Anchor is usually more price elastic than attach Never discount your price on attaches
Predictive Analytics
Principle Component Analysis Problem/Issue
Your price can be affected by many factors PCA is a variables reduction process Always prioritize the most influenced
factors
Predictive Analytics
Principle Component Analysis Analysis
Predictive Analytics
Correlation matrix SAS Factor procedure Excel Data Mining Add In
Principle Component Analysis Lesson Learned
Principle component analysis is not factor/cluster analysis Must eliminate any possible outliers prior to run PCA Always better to run PCA before predicting pricing impact
Predictive Analytics
Margin Bridge Analysis Problem/Issue
A way to analyze pricing/margin changes impacted by various mixes
Predictive Analytics
$5,000
-$2,000
-$1,500
-$1,350 -$800
$1,500
$2,000
$3,000 $5,850
Year-2015 Pricing Costs Segment Mix Channel Mix Product Mix SalesEnhancement
New Product Year-2016
Margin Bridge Analysis Analysis
Predictive Analytics
It is not Price-Volume-Profit Analysis in accounting and it can not be achieved by using simple algebra Multiple Linear Regression involved (GLM) and some assumptions are made
Margin Bridge Analysis Lesson Learned
Time consuming to transform the data and test the hypotheses More than likely the results are directional rather than solid science
Predictive Analytics
Classification and Decision Tree Problem/Issue
Regression trees for prediction Quickly identify price sensitive segments Systematically search for low-hanging
fruits Evaluate existing pricing actions Classify potential targets using the data
from prior promotions
Predictive Analytics
Classification and Decision Tree Analysis
Predictive Analytics
Decision Tree based Methods Rule-based Methods Memory based reasoning Neural Networks Naïve Bayes and Bayesian Belief Networks Support Vector Machines
Classification and Decision Tree Lesson Learned
Require a large amount of data Easily understandable and transparent Great visualization of representation However, trees must be pruned to avoid over-fitting of the training data
Predictive Analytics
QUESTIONS?
PRICING ANALYTICS