Travelers Cheque (TC) & Gift Cheque (GC) Database - Direct Marketing Strategy: Update on key learnings and next steps

Transcription

1 Travelers Cheque (TC) & Gift Cheque (GC) Database - Direct Marketing Strategy: Update on key learnings and next steps 1

2 Background existing TC customers/ Amex CardMembers (CMs) prospects for TC/ Amex CMs Prospects for GC=>Mailing dropped on weeks 12/2 & 12/8 Key strategies were: Expand upon the potential of existing customers through a refinement/ improvement of our profit scoring/ calculation Objective: Concentrate even more on the best customers & generate even more sales & profit/mailed customer Approach: Test of a new CRM driven profit calculation for customer profitability on TC Direct Sales customers Leveraging more systematically the Amex Card prospects potential (15 Million CMs) through the development of CRM driven Profit prediction and Response prediction models for TC & GC Objective: Ability to identify key demographic or Amex Card data that are statistically predictive respectively of Profit and Response in order to concentrate the solicitation of the highest profit potential and highest response potential Amex Card prospects for TC & GC Approach: test new TC & GC profit & response models for TC & GC on Amex Card prospects 2

3 Background (cont.) Start investigating new creative paths for GC Objective: test a message more in line with general positioning Approach: Test a new Postcard against the Q4 postcard format 3

4 TC sales on Existing Customers/Results Existing customers are our BEST target so far for Direct Marketing with continue performance improvement through a higher level of sophistication in our targeting & profit score Our successive refinements in the profit calculation for existing customers show a clear positive impact on the Response Rate! $ 2 Million Sales with only 1300 best customers Response Rate 20.00% 18.00% 16.00% 14.00% 12.00% 10.00% 8.00% 6.00% 4.00% 2.00% 0.00% Q4 Year 1/No mail test Q4 Year 1 test/profit score reference Q4 Year 2 test/profit score reference Q4 Year 2 test/refined Profit score 4

5 TC sales on Existing Customers/Results (cont.) While we improved substantially the response rate, we were also able to maintain a high order size which is 140% above the average order size in the US. TC Order size $2,000 $1,800 $1,600 $1,400 $1,200 $1,000 $800 $600 $400 $200 $0 Average Market level Q4 1999/No mail test Q4 Year 1 test/profit score reference Q4 Year 2 test/profit score reference Q4 Year 2 test/refined Profit score 5

6 TC sales on Existing Customers/Results (cont.) In line with this evolution, DM on existing TC customers is able to deliver positive ROI ROI 450% 400% 350% 300% 250% 200% 150% 100% 50% 0% Q4 Year 1 Q4 Year 2 test/profit score reference Q4 Year 2 test/refined Profit score 6

7 TC sales on Existing Customers/Results (cont.) The treatment of the response through our phone center offers clear advantages with the X sell opportunities it provides A Direct Sales Consumer buys 2 products in average/order 7

8 Traveler Cheque Mystic (Trigger-Based) Pilot/ Key learnings & conclusions 8

9 Background (cont.) Mystic is an internal Amex Card database where among other data, airline tickets purchases with an Amex Card are stored with info on airport destination & departure date CMs mailed/ CMs not mailed as control Drops between July & December (Mailing in home 2-3 weeks before departure date) Our key strategies were: Expand on past test results by refining the targeting on key profit variables in order to better leverage the natural response lift generated by a travel trigger Objective: Focus on key TC profitability variables in order to boost the response rate and the profit contribution and over compensate the additional costs generated by small successive DM waves implied by a trigger driven approach Approach: Add & test an additional 55+ age selection criteria on the top of Domestic travel with 1000 miles distance at least & all international travel 9

10 Background (cont.) Fine tune the best/most cost effective creative format Objective: Define what will be the most cost effective contact format Approach: test letter vs postcard vs Start investigating additional offers Objective: Assess if an additional offer can boost the response rate sufficiently to generate additional profitable sales Approach: test additional 55+ book offer vs no additional offer 10

11 Traveler Cheque Mystic Pilot/ Results A travel trigger like Mystic is NATURALLY a very powerful response booster for Traveler Cheque Our tests between Year 1 & Year 2 indicated that even with no mailing, the travel trigger incited more people to buy TC naturally! 1.60% 1.40% Response Rate 1.20% 1.00% 0.80% 0.60% 0.40% 0.20% 0.00% No triggers target/no m ailing Mystic target Year 1/no m ailing Mystic target Year 2/no m ailing 11

12 Traveler Cheque Mystic Pilot/ Results (cont.) An 55+ additional selection criteria helps boost marginally the response rate and the order size indicating that the trigger effect is the main driver of response For the response rate, a 55+ age criteria achieves better results! Response Rate 1.90% 1.70% 1.50% 1.30% 1.10% 0.90% 0.70% 0.50% 0.30% 0.10% Year 2 Mystic target/no 55+ criteria Year 2 Mystic target/55+ criteria 12

13 Traveler Cheque Mystic Pilot/ Results (cont.) A mailing activity on a travel trigger list like Mystic effectively boosts even more the response rate for Traveler Cheque Response Rate 3.75% 3.50% 3.25% 3.00% 2.75% 2.50% 2.25% 2.00% 1.75% 1.50% 1.25% 1.00% 0.75% 0.50% 0.25% 0.00% No trigger/no mailing Year 1 Mystic target/no mailing Year 2 Mystic target/no mailing Year 3 Mystic target/m ailing 13

14 Traveler Cheque Mystic Pilot/ Results (cont.) While potentially more cost effective, a Post Card format looses 25% response efficiency and an format looses 61% response efficiency compared to a letter format 4.00% 3.50% Response Rate 3.00% 2.50% 2.00% 1.50% 1.00% 0.50% 0.00% Year 2 Mystic target/letter Year 2 Mystic target/postcard Year 2 Mystic target/ (*) (*) Web Site was down after 1 month of activity 14

15 Traveler Cheque Mystic Pilot/ Results (cont.) The new Book offer seems to be working for Mystic on a 55+ target but still in line with the traditional offer! 3.00% Response Rate 2.50% 2.00% 1.50% 1.00% 0.50% Year 2 Mystic target/offer reference Year 2 Mystic target/new Book offer 15

16 Traveler Cheque Mystic Pilot/ Results (cont.) By having a clear boost on the Response Rate, the mystic test was able to generate positive ROI on the letter & Post Card format % % % % ROI % % % 50.00% 0.00% % % Mystic/no m ail Mystic/Letter Mystic/Postcard Mystic/Em ail Note: ROI are roll out scenario based 16

17 Traveler Cheque Mystic Pilot/ Results (cont.) E mail format could become profitable if we were able to leverage a better cost/ ROI 300% 250% 200% 150% 100% 50% 0% -50% -100% Mystic/Em ail/actual costs 10 c per Mystic Em ail/projected cost 2.5c per 17

18 New Profit & Response Models/Initial findings 18

19 TC models objective & methodology Objective : utilize statistical methods to achieve a higher dollar yield on a marketing campaign for the Traveler Cheque product. By contacting fewer, but more profitable & responsive Amex CMs, the average PTI per dollar of acquisition cost should be significantly higher than contacting a random subset of the Amex Card base or a subset generated by rules of thumb. Methodology: the analysis driven by CIM was based on the Q4 pilot responders in order to extract the most significant variables from the Amex CM database that are explaining respectively profitability and response for TC purchase and then use those same variables to select who to contact for TC prospecting within the card base. 19

20 Is the new TC profit model working? While Amex CMs purchasing TC at our Call Center are virtually always profitable, the new TC profit model identifies very effectively the best TC customers among CMs. 76% of total profit Integrate unprofitable Customers in the mix Mean profit per customer Customers Groups 26% of total profit 20

21 What are the database variables that are statistically predicting profit for TC? 1st: Foreign spending on the Card (the more...the better) 2nd: Age (the older...the better) 3rd: Platinum Card ownership 4th: Gold Card ownership 5th: Number of charges on the Card on High level industries (the less...the better) Gender: Male is better than Female Credit score (the higher the better) NB: New compared to Q analysis 21

22 What is the average profile of the top 20% profitable TC purchaser? High Foreign spending on the card (above $4000/year) Age 55+ Platinum or Gold CM Male Very High Credit score 10+ Charges on the Card on (High level) industries => 3 Million CMs within target for Top 20% => 7.5 Million CMs within target for Top 50% 22

23 Is the new TC response model working? The new TC response model identifies very effectively the best TC responders among CMs. 74% of total response Mean Response per customers group 4% 3% 3% 2% 2% 1% 1% 0% Customers Groups 23% of total response 23

24 What are the database variables that are statistically predicting response? 1st: Credit Score (the more...the better) 2nd: Number of travel destinations over last 12 months(the more...the better) 3rd: Amex Fees amount (the less the better) 4th: Number of charges on SE expansion (the less...the better) Enrolment in Amex Luggage Insurance protection 5th: Restaurant spending on the Card (the less the better) 6th: Enrolment in Avis preferred program 7th: Monthly mortgage payment (the higher the better) NB: New compared to Q analysis 24

25 What is the average profile of the top 20% profitable TC purchaser? Very high Credit score At least 2.26 different international travel destinations/year Use less additional services from Amex High spender in SE expansion rocs Enrolled in Luggage Insurance Low spender in restaurant with Amex Card Enrolled in Avis Preferred pgm Higher monthly mortgage payment => 3 Million CMs within target for Top 20% => 7.5 Million CMs within target for Top 50% 25