Introducing: Automated Personalization with Dynamic Yield’s Predictive Targeting

Personalization has been gaining popularity for some time now, driving businesses to invest in a more individualized approach. Many have scraped together what they feel is a good-enough programmatic targeting setup, leaving them to believe personalization has been properly put in place and that results are imminent.

But as marketers have quickly learned, true personalization is much more involved, requiring a lot of time, effort, and resources. Properly executed data-driven personalization relies on segmented website traffic (in itself a tedious data-heavy task), numerous test deployments with conclusive results, analyzed data, and measurement of every tested variation against each audience segment in order to determine optimal programmatic targeting rules.

Further evidence that marketers could use assistance is seen in a recent Personalization Maturity Assessment survey which found only 17% of companies claim to have personalization in place, citing “lack of resources” as their leading barrier (49%). These difficulties inherent to personalization often drive marketers to resort to guesswork or intuition-based targeting, leading to suboptimal results, for obvious reasons.

But while personalization is evidently still a nascent effort for many brands, A/B testing is a widespread and relatively easy-to-execute practice among most of today’s online businesses, continuing to faithfully identify the best variation or experience among your website’s traffic. However, the actions taken following a test are intrinsically flawed, as marketers tend to serve the winning variation to an entire visitor pool, effectively deploying a “winner takes all” strategy, the opposite of a personalized approach.

While an A/B test may reveal the best variation across your traffic on average, there will always be segments of visitors for whom the winning variation is not optimal. In other words, “winner takes all” may be the best bet for the majority of your visitors, but you’ll be compromising the experience for the other portion of your visitors, and most likely, leaving money on the table.

Introducing Dynamic Yield’s Predictive Targeting

Combining the best of both A/B testing and personalization, Predictive Targeting, a new AI-Powered solution by Dynamic Yield, allows businesses to automatically convert every A/B test into an effective personalization opportunity. For those well-versed in the land of A/B testing, Predictive Targeting is a convenient bridge and entry point into the world of personalization.

Predictive Targeting runs in the background of every experience launched through Dynamic Yield, analyzing the performance of each variation across every traffic segment in real-time and identifying opportunities to serve the most relevant content to select audience segments. Once a personalization opportunity is uncovered, marketers have the flexibility to take immediate action, deploying the necessary targeting setup predicted to drive the most uplift in revenue for each segment, with a single click.

Let’s examine the following use case:

In an effort to identify the optimal product recommendations strategy to deploy on a PDP, a merchandiser from an iconic eCommerce sports site launched a test with three variations, each receiving an equal portion of traffic:

Variation A

1/3 of the traffic received recommendations for products that were manually selected (control group).

Variation B

1/3 of the traffic received recommendations for products that were similar to the product in focus.

Variation C

1/3 of the traffic received recommendations based on a fusion of strategies, including both similar and complementary products.

The results come back, stating conclusively that Variation C is the winner, expected to generate a 34.4% uplift in conversion rate vs. the control group. Following common practice, that merchandiser would intend to deploy the variation that on average expected to perform best overall for all users. In this case, Variation C. But would doing so really generate the best performance for all users?

During this test, Predictive Targeting ran in the background, analyzing the performance of each variation across every one of the site’s traffic segments. It revealed that if Variation C was served to the Audience titled “Frequent Visitors” while all other visitors received Variation B, a whopping uplift of 54.1% would be generated overall. And so it was.

Another customer recently launched a simple test to determine the best copy for a Display Ad campaign, consisting of a control group and treatment variation. The results revealed the control group to be the unequivocal winner while the treatment significantly hurt performance. Disheartened by the results, the customer was faced with an automated personalization recommendation. Instead of discarding the treatment, Predictive Targeting advised the user to serve the treatment to a subset of ad viewers (30% of traffic) who was identified to actually perform better to the treatment variation. In this example, Predictive Targeting effectively converted a “losing” test into an opportunity for better personalization!

With Predictive Targeting, you can overcome familiar marketing limitations, as it automatically handles all of the tedious data analysis tasks behind the scenes so that you won’t have to. All Predictive Targeting suggestions are statistically significant and data-driven, ridding marketers of the need to rely on guesswork or heavy duty analysis when executing personalization strategies.

Predictive Targeting does not require any setup or development and is automatically ingrained in any campaign and test launched through Dynamic Yield. That means across any channel or platform, including mobile apps, email, and offsite advertising, allowing you to leverage all of your user data, surfacing recommendations at the moment they are identified.

With Dynamic Yield’s Predictive Targeting you receive:

• Constantly running algorithms for all campaigns for identification of revenue uplift opportunities
• Real-time recommendations delivered through the platform and by email for the highest impact opportunities
• 1-click deployment offering full control and flexibility to deploy any recommendation
• Winner declaration per audience segment vs. the winner takes all approach

Predictive Targeting is available to all Dynamic Yield customers. During its beta program, Predictive Targeting was fully tested and vetted, crunching data for over 2,500 experiments in real time and generating successful personalization strategies and deployments for some of the world’s leading retailers. With Predictive Targeting, personalization has become so much easier.