Perhaps, the least discussed, least understood and possibly the most valuable form of marketing analysis are the multivariate techniques. Most marketing professionals emphasize and discuss A/B testing. It is applied to many outbound and inbound marketing tactics such as email subject lines, landing page titles, call to action text and buttons and various marketing automation campaigns. However, A/B testing does not address the more complicated marketing questions where the overall approach to audience receptivity is called into question. Understanding whether the campaign is approaching the target audience in the best way to maximize interactions and conversion rates requires more than changing a single line or heading. Multivariate analysis is the best way to achieve this objective.
Analyzing multivariate campaigns is not a simple process. The manipulation of multiple variables and measuring the effectiveness of the change requires a sample size sufficient to measure the number of variables or conditions applied to the campaign. It also requires the marketer to group the responses into meaningful categories that can be used to determine which alternative(s) are most effective.
Multivariate testing is often used in the design of websites. Together, the content, font, graphics, tone and layout of a webpage represent the brand and branding for a company. As a group the communicate the brand to visitors. A/B testing is not efficient when so many variables need to be tracked and the response is not “yes”, or “no”. Responses to webpage designs consider emotion, effectiveness of communication, tome of the information, casual verses professional, and on and on. Ultimately, the winning design is chosen because it delivers the best experience for the visitor and generates the best response. A/B Testing is performed on the winning design to incrementally improve its performance.
Multivariate testing can also be applied to email content, sales promotions, company collateral, surveys and much more. There are many forms of analysis with some more applicable than others depending upon the information collected, concerns to be addressed and database size. Wikipedia lists 17 different statistical models and while I like and trust statistics and quantifiable information, 17 is a lot.
Consider instead evaluating which techniques will be most effective based upon the available data and familiarity with specific analysis methods. It is very important to vary the type of analysis as there is no one silver bullet. Either one or more form of analysis such as multiple regression, discriminant, conjoint, factor or cluster analysis may be appropriate. Experience and skill are needed to make the best selection(s).
As this is clearly a subject too broad for a blog, might I suggest “Multivariate Data Analysis for Dummies” by Brad Swarbrick. It’s a good read. And for software to assist, I suggest evaluating the top rated Multivariate analysis products on Capterra.
In summary, don’t think of multivariate and A/B testing as opposites or that one should be favored over the other. They are complementary with each having a role to play on a successful marketing team. Each contributes to Marketing Excellence.
Leave a comment