Presentation on theme: "Predictive Modeling for Marketing"— Presentation transcript:
1 Predictive Modeling for E-Mail Marketing Arthur Middleton Hughes – Senior StrategistAnna Lu - Director of Research and AnalyticsPredictive Analytics World Feb 18, 2009
2 What Does E-mail Marketing Do? 3/25/2017Produces online sales – in many casesProduces retail sales – in many more casesProduces customer retention and loyaltyHelps to acquire new customersAnnounces new productsCreates cross-sales and upgradesCan be the most powerful and cost effective marketing method that marketers have available today -- particularly in an economic downturn
3 E-mail’s Role Not Understood 3/25/2017In many companies, is not recognized as the marketing powerhouse that it isIt is somewhere off on the side, producing Web sales which are about 3% or less of total salesThat may be the perception, but companies that think that way are missing the boatHere is the reality…
4 E-mail Produces Four Times as Much Offline as Online 3/25/2017
5 The value of multi-channel customers 3/25/2017marketing budgets are often based only on online salesThis is a mistake, because produces four times as many sales offline as they do onlineCalculate the true effect of so that the marketing budget can reflect the true worth of e- mail marketing
7 Predictive models seldom used Most marketers today do not use predictive modeling. Why not?Predictive modeling is used in Direct Mail where the CPM is $600 or more. In marketing the CPM is $8 or less. Many marketers feel that the savings from a model would not pay for the model.Many marketers are young people who have never heard of predictive modelingThe philosophy is: “Mail ‘em all. Someone is going to buy…”This attitude is beginning to change. Here’s why….
9 People are unsubscribing It costs between $10 and $40 to acquire a permission based subscriber address.Inboxes today are so crowded with s that millions unsubscribe or delete s en masse without reading them.A relevant to a good customer gets lost in the spam.Many marketers are mailing too oftenThe annual loss from unsubscribers from large mailers comes to millions of dollars
10 Predicting the unsubscribers Unsubscribe rates are often 3% or more per month.If a mailer has 4 million subscribers, and the value of each subscriber is $15, he could be losing $21 million per year.If the unsubscribe rate could be reduced by 10% he would save $2.1 million per year.You could pay for several predictive models with that kind of saving.
11 Finding Likely Unsubs with CHAID Case Study: Loyalty program for a major US low cost airline
12 Program BackgroundFrequent flyer program for a major low cost airline in USSemi-weekly program offered to members who wish to accumulate "points" they can put towards flights, SkyMall products and moredrives a significant percentage of the total revenue
13 Business Problem18.5% have opted out of the program communications but they represent 30% of the total revenue generated by all membersIn addition, 88% of the opt-outs happened within the past 12 monthsStatus% of Program Base% of Revenue Generated (Lifetime)% of Revenue (2 yrs.)% of Revenue (12 months)Mailable81.5%70.6%78.7%82.4%Opt-out18.5%29.4%21.3%17.6%
14 Objective Understand key characteristics of previous opt-outs Identify likely unsubsInitiate save programs to prevent unsubs from happening
15 Analysis Background Random sample of 5% of member base Approx 50 predictor variablesProgram attributes such as enrollment date, mile accumulation, usage, recency of mile redemption, total reward points, Lifetime revenue, etc.behaviors such as opens, clicks and purchases (from s sent)Response variable – Unsubscribed versus still mailable (binary level variable)CHAID (Chi-square Automatic Interaction Detector) algorithmCross validation method
16 About CHAID A type of decision tree technique Use of the chi-square test for contingency tables to decide which variables are of maximal importance for classificationAdvantages are that its output is highly visual and easy to interpretOften used as an exploratory technique and is an alternative to multiple regression
17 Output (Partial) % Unsub Overall % Unsub among people with # of opens in last 60 days=1
18 Predictors Selected # of e-mails opened last 60 days Days since loyalty club enrollment# of s opened last 30 days# of Bonus (partner) credits earned YTDDays since last travelDays since most recent opened or clickedDate of Last earn/ or redemption of flight/ or Bonus (partner) credit# of s opened last 365 days# of vouchers redeemed in lifetime
19 Node GainGain Chart on model development sample
20 RevenueTop 10% of the members contributed to 67% of total revenue
21 X-Tab: Node vs. RevenueEach of the top nodes have high revenue producing members
22 Identifying most profitable flyers 4% (or 120K) of frequent flyers contributed 15% (~$3.1 million) of program revenue
24 Using the output of the model Now that you know those most likely to unsubscribeAnd know who are the most valuableYou can single out these folks and make them an offer that they cannot refuse.Analytics helps the airline target the right people.
25 How modeling reduced churn In one year, analytics was used for a wireless phone company –Cingular - to reduce monthly churn by 26% -- Millions of dollars.
26 Identify Best-Customer Look-Alike with Logistic Regression Case Study: US off-price e-tailer
27 BackgroundOff-price e-tailer of name-brand apparel and other goods in USis their single largest marketing channel, and their most important retention toolcommunication delivers 40% of the total revenue
28 What can be measured Attrition and retention Migration upward and downwardIncremental sales per program and per seasonFrequency of seasonal purchasesDollars spent per trip and per seasonNumber of departments shopped per trip and per season.Number of items shopped per trip and per season–Share of customers’ wallet
29 Business ProblemAbout 50% of revenue are actually driven by their loyalty club membersAn annual membership fee is requiredSize of loyalty club is small – just 1.8% of baseClient asked:Who should we focus as the next tier of subscribers amongst the other ~98% of the listWho look like the best customers I haveHow can we find people who might become best customers if nurtured
30 Objective Understand what variables describe best customers Identify likely best customersInitiate programs to nurture these subscribers, to keep them happy
31 Analysis Background Random sample of 10% of e-mail subscriber base Approx 10 predictor variablesAttributes such as # of lifetime purchases, first/most recent order, address acquisition source, etc.behaviors such as tenure, opens, clicks and purchases (from s sent)Response variable – Loyalty program member vs. non-Loyalty program member (binary level variable)Logistic RegressionCross validation method
32 About Logistic Regression Prediction of the probability of occurrence of an event by fitting data to a logistic curveVery useful techniques when one wants to understand or to predict the effect of a series of variables on a binary response variable (a variable which can take only two values, 0/1 or Yes/no, for example)For example, it’s help to anticipate the likelihood of customers responding to a direct mail, or the likelihood a person is about to churn from a subscription
33 Impact of Predictors Some variables used included: Total # of purchasesThe more the betterTime on fileThe younger the betterMonths since first purchaseMonths since last purchaseThe less (or more recent) the betterTotal s clicked on over the past yearTotal s opened over the past yearThe more the better… though not always predictive
34 Model GainGain Chart on model development sample
35 Now that we know who to target… The model enables us to focus on those most likely to be interested in the loyalty club.We can target only those folks with messages and rewards that will get them to join.We make them offers that we could not afford to offer to everyone.How the model boosts profits and reduces churn…
36 Model beats random select A model predicts those subscribers who would be interested in a particular product.Mailing these 273,334 produces 842 sales and only 273 unsubscribers.If the model had not been used, there would have been only 41 sales and 3,553 unsubscribers.Replacing each unsubscriber costs $14.Without the model, the mailing would have been a disaster.
37 ConclusionsPredictive modeling is just getting started in marketing.Reason: s are so inexpensive that the attitude was: “Blast ‘em all!”We now realize that subscribers are very valuable. We can lose them by random blasting.Models help us by reducing unsubscribes and also by identifying those subscribers who are most interested in what we have to say.Predictive modeling works with marketing.
38 To learn more….Available from Amazon.com or BarnesandNoble.com
39 Thank you for viewing.For more information, please contact: Arthur Middleton Hughes, Senior Strategist | Anna Lu, Director of Research and Analytics |