Readerís Digest Canada typically sees response rates that are one or two percentage points higher than the industry average for their catalog mailings of CDs, DVDs and books. But with better data mining technology and digital print capabilities, Mathieu Peloquin, VP Marketing, Product Lines, and his team were intrigued with the notion of customizing their print catalogs based on past buying data.
ďWe wanted to remove the barrier of Ďone-catalog-for-all.í We wanted to test fully customized catalogs and determine how the variable model can impact the customer,Ē he says.
Peloquin wondered if they could increase sales enough to pay for higher printing costs by using past purchase information to create a truly personalized catalog for each customer. Could they make their catalog merchandizing operations act more like an analytics-driven ecommerce site?CAMPAIGN
To begin, they set up three tests of varying catalog customization involving 47,103 customers who had purchased an item in the past 10 weeks. The number of consumers receiving each mailing would be different, as it was determined by how many fit the varying criteria for each test.
Because Peloquin and his team wanted to find the best ROI scenario, they tested catalog lengths of 16, 20 and 24 pages, as well as correlating, full-color self-adhesive sweepstakes product stamp sheets against versions without.
Moving forward, all three tests were measured against a control group of 15,701 who received a typical version of their non-personalized promotion.
The control mailing included:
o A 24-page, 5-1/4-by-8-1/4-inch booklet with information on 43 best-selling CDs
o A sweepstakes package with a letter/certificate that included the consumer's personal information in black type.
They used only music items for the control group because that category had the widest margins and Peloquin wanted to establish a hard-to-beat ROI benchmark.
The tests involved an algorithm system that produced numerous product-affinity combinations and patterns. As a result, each test included a catalog populated with four personalized products based on most-recent purchase and/or affinity patterns (items bought by other consumers).
Peloquin didnít want any changes in design to affect the findings, so they kept the same catalog layout for the control and test packages.
Hereís how they modified the test catalogs:
- Test #1A. Customized catalog, but no cross-sell
Peloquin wanted one package to help them see what kind of variations could be achieved without going too crazy with customization. The catalogs, which mailed to 10,468 consumers, used a handful of personalized touches.
o A 16-, 20- or 24-page booklet/catalog, with size depending on the number of products relevant to the customerís history
o One product line -- either CDs, DVDs or books -- based on the customerís most recent purchase
o Due to affinity purchase patterns dialed up by the systemís algorithms, products were placed inside the catalog in sequence of relevance
o A full sweepstakes package, including a personalized letter, prize offer and stamp sheets
o Product images in the catalog and on the stamp sheet were sequenced to relevance
o All items were printed in four color
- Test #1B. Simplified catalog version
Notable differences for this version included:
o No sweepstake elements at all -- and, therefore, much less personalization
o Volume of the mail drop: 5,234
- Test #2. Customized package/cross-sell multiple products
This test allowed Peloquin and his team to create a package with hundreds of merchandising combinations that went to 15,700 consumers.
o Products based on previous purchases and an affinity pattern, but included items from all three product lines (CDs, DVDs, books)
o Items were sequenced based on affinity patterns
o A full sweepstakes package, including personalized letter, prize offer and stamp sheets
o The product images in the catalog and on the stamp sheet were sequenced to relevance
o Varying stamp sheet length based on page count
o The size of the catalog -- 16, 20 or 24 pages -- was decided by the number of products relevant to the customerís history
o All items were personalized in four-color process
The results proved that even an old direct marketer like Readerís Digest can learn new DM tricks. Indeed, the findings gave Peloquin and his team overwhelming reason to move away from the concept of ďone-catalog-for-all.Ē Overall, their test results saw average sales lifts over the control package: 49% for Test #1A, 16% for Test #1B and 74% for Test #2.
Here's how the response data broke down by product category compared to the control package:
Test #1A: Customized catalog without cross-sell (and one product line per recipient):
o CD buyers, 27%
o DVD buyers, 111%
o Book buyers, 45%
Test #1B: Simplified version (and one product line per recipient):
o CD buyers, 7%
o DVD buyers, 68%
o Book buyers, 9%
Test #2: Customized package/cross-sell multiple products (multiple product line offers):
o CD buyers, 35%
o DVD buyers, 78%
o Book buyers, 83%
Peloquin and his team also received very encouraging anecdotal feedback when it came to all three tests. ďWe heard comments like, ĎI could have bought every item in the book.í So, those products really resonated.Ē
Going from a system of creating one catalog for all vs a digital template where customer behavior fills in the key spots wasnít easy for Peloquinís design team, but the results prove that itís worth it to test. They still need to conduct another test or two to determine the exact size of the catalog. ďWe were near our ROI requirement. We are only months away from rolling out.Ē Useful links related to this article
Creative samples from Reader's Digest Canada's catalog test:
Xerox 1:1 Lab - provided the technology for the variable catalog tests:
Terminal Van Gogh - handled 1-to-1 marketing consultancy and solutions for the tests:
Exstream Software: - provided additional technology for the tests:
Reader's Digest Canada: