Testing How Personalized Content Improves Customer Experience [#CZLNY]

In theory, we all know that testing and personalization are important ways to extract as much engagement and business value from our marketing. However, many of us still don’t invest in them. At ClickZ Live today in New York City, Jake Sapirstein of SAP shared some learnings from a two-year testing program to learn the impact of personalization on landing pages.

“All the energy, work, and media spend in marketing are driving impressions to a digital landscape,” he said. “It’s important to spend a fraction of that budget, and some of the energy, to optimize the landscape itself.”

SAP focused on the initial interaction and landing page, starting with keyword-driven targeting, or keyword insertion. Simply changing landing page content on sap.com based on the terms that were used in search to reach that page had significant results. “We tested for lead generation – not just impressions – in multiple locations and different cultures.” he said. “We saw consistent results with an aggregate of 70 percent lift in registrations.”

About a year later in 2014, SAP embarked on a more sophisticated set of personalization tests. A series of A/B tests were done on the U.S. home page of sap.com of generic content as control and challenger pages that were personalized by industry. The personalization was done via Demandbase, which allowed SAP to detect company-based information per visitor and serve content based on the firmgraphic, in real time. Traffic was split using the Adobe Target platform, where half got generic and half got the industry personalized content.

So why should brands test?

  1. To validate or disprove hypotheses about what works best
  2. Digital content testing means we can fail fast, and move forward with more conviction
  3. Testing uncovers precise impacts of a specific change or tactic

The results of SAP’s tests were mixed. “We never assume; that is why we test,” Sapirstein said. “We got consistently good results in at least one industry – banking and a few others – but others did not do as well, meaning we did not get a lift in response due to the personalization.”

A similar test was done in Germany. Results were similar, where the personalized pages only surpassed the generic pages in certain industries, like banking, which was 214 percent higher.

“If we knew everything there would be no testing industry,” Sapirstein said. “SAP has found that the marketing team is only right on hypotheses about half the time. Nearly 50 percent of our tests validate, about 30 percent are bakeoffs (and need more testing to prove a winner), and 20 percent are surprises, where the hypothesis was disproved.”

Related reading

Glittering cloud with electric neurons
Businessman holding a smartphone, with a visual representation of cloud computing, a little white cloud connected to various icons, hovering above it.