CRM - May 2008 - (Page I-10)

10 May 2008 Sponsored Content Multivariate Testing: A Formula for Direct Marketing Success Initiating, nurturing, and growing any kind of relationship takes hard work, dedication – and a little luck. But when you’re talking about a customer relationship, luck’s not going to cut it. You need the right tools, strategies, and resources to positively impact customer behavior across the lifecycle. Historically, marketers have turned to A/B testing (e.g. champion/challenger) to glean insight into their customer relationships, while combating the challenges of generating additional revenue, decreasing costs, and demonstrating ROI. Although A/B testing can provide insight to accomplish these goals, it can often be a lengthy process, expensive, and too narrow in scope to provide a meaningful impact on your direct marketing campaigns. THE SOLUTION DEFINE TEST OBJECTIVES It is important to determine what you’re looking to accomplish prior to conducting any test. Do you want to increase response rates to drive customer acquisition? Are you looking to build customer loyalty? Or, do you want to increase revenue while decreasing marketing costs? Setting clear objectives at the beginning of the project will ensure the test is designed properly and the results are truly relevant to your hypothesis. DETERMINE TEST VARIABLES AND SAMPLE SIZE Sample Multivariate Test Results Test Variables Creative Copy Timing Buck Slip Outer Envelope Pre-Email Free Gift Discount Teaser Message Testimonial Response Rates Control 1.12% 1.02% 0.96% 0.72% Test Difference 0.60% 0.54% -0.48% 0.33% -0.30% -0.30% 0.50% 0.06% -0.07% 0.06% 1.72% 1.56% 0.48% Statistically Significant Variables (highlighted variables are the optimal response rate package) 0.67% 0.51% 0.35% 0.23% 0.17% 0.08% 1.05% 0.37% 0.21% 0.85% 0.29% 0.10% 0.14% No Statistical Significance Package Control Response Rate – 5.35% Optimal Response Rate Package – 7.32% Multivariate testing is a statistical method that combines multiple ideas such as creative, promotional materials, advertising, offers, and price into one comprehensive direct marketing test. Unlike traditional methods, you can now test multiple variables simultaneously to determine the optimal contact strategy. What’s more, you can also isolate the impact of each individual variable as well as understand the interaction between multiple variables to identify the optimal response rate package – not just an iteratively better package. Multivariate testing also eliminates variability in serial A/B testing from factors such as seasonality, deliverability, audience, and offer. In short, by testing multiple variables within one test, you can: • REDUCE TESTING TIME AND SAMPLE SIZE • INCREASE FLEXIBILITY OF TEST DESIGNS • INCREASE EFFICIENCY • EXPAND DEPTH OF INSIGHT • SHORTEN TURNAROUND TIME FROM KEY LEARNINGS TO EXECUTION While multivariate testing offers the flexibility to test up to 36 ideas, it can be costly if not appropriately structured. When determining which variables you want to test, it is important to remember that they must be independent, bold, and test worthy. The goal is to isolate only those variables that can provide truly actionable information. It’s also important to determine the appropriate sample size, which requires both creativity and scientific discipline. Despite the fact that more variables are being tested, the sample size for a multivariate test isn’t significantly larger than that required for a traditional A/B test. DESIGN AND EXECUTE TEST variables) has the greatest effect on response rate – and of equal importance – which ones that don’t. DOES IT WORK? Through our suite of data analytics services, ACCENT has strategically guided a number of clients through the design and execution of multivariate tests with impressive results. In fact, for one of our clients, the insights gained from the execution of a single multivariate test yielded a 400% return on investment – significantly increasing response rates, which added millions of dollars to the top and bottom line. ABOUT ACCENT ACCENT is an international provider of customer lifecycle management solutions. We partner with world-class companies to help them acquire, support, retain, and grow their customers at every stage in the customer lifecycle. Through our suite of data analytics, direct marketing, contact management, and fulfillment services we deliver high quality, integrated marketing solutions that grow valuable client and customer relationships. Founded in 1993 and majority owned by MDC Partners, Inc., ACCENT employs more than 4,000 people across the United States, Jamaica, and the Philippines. For more information, call 866.623.0018 or visit www.accentonline.com. Once you have determined the number of variables you want to test, the analyst can then create a test matrix, which consists of scientifically defined “formulas.” Each formula is like a single test element, providing one unique piece of information about each variable in the test. Once complete, the test matrix becomes your roadmap for writing copy, designing creative, and defining measurement criteria for each test campaign. MEASURE RESULTS Before conducting any multivariate test, consider the following: Unlike A/B testing, multivariate testing measures variable (test element) response rates – not ‘package’ response rates. Isolating the effects of each variable will allow you to determine which variable (or combination of www.accentonline.com

Table of Contents Feed for the Digital Edition of CRM - May 2008

CRM - May 2008 Contents Front Office Feedback Reality Check Customer Centricity The Tipping Point Is CRM Too Hard for Microsoft Vendors Go Virtual For Feedback Sense-sational Marketing How UGC Can Benefit CRM DestinationCRM Dashboard Price Check, Aisle 5 Required Reading The Moving Target The Excellence Myth Seven Steps to SOA Success And They're Off! Are You Ready to Party? Skin in the Game The Right Numbers Secret of My Success Re: Tooling Connect Pint of View

CRM - May 2008

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.