Putting a user-friendly front-end on Statistics
A/B testing platform Optimizely gets funding from some big names
The world runs on numbers, which can be confusing when a company runs on words.
Optimizely is bridging that gap one app at a time. This week, the firm announced the release of Optimizely Personalization, a user-friendly front end that helps analyze and sharpen the user experience when visiting a web site. The solution allows users to construct and alter their own retail web sites, tagging each element to track how users interact with the site, as well as linking with past data on user purchases.
What is Personalization? “When successful within the web site or app, it provides you with more relevant information,” explained Jon Noronha, senior product manager at Optimizely. What it should not be is creepy or pushy. “There are cases when personalization chases you around.”
Easy as A/B
Optimizely Personalization has a heritage well-rooted in A/B Testing, the practice by which two versions of the same message are tested side-by-side to produce a measured outcome for comparison. The more effective version gets adopted.
Companies are sitting on troves of data. But they are challenged trying to understand the groups that are contained within the data, and they are unsure where to focus, Noronha explained.
Personalization starts by linking data, using a visual interface that allows the user to tag elements on the web page to track what people click on. That information can be linked to previous transactions. There are also about 40 APIs built into personalization that can also link to other analytic tools, like Salesforce for CRM or Google Analytics, or Skymosity, which can factor in weather conditions as it affects sales.
The A/B testing feature embedded in Personalization will automatically skim off five percent of customer traffic and send them to a generic online sale web site, while sending the bulk of the traffic to the redesigned web site. Data from the two can then be compared to see if the changes produced any meaningful positive impact, Noronha said.
There is also a machine learning capability embedded in Personalization that will help users identify distinct segments in their data. Such group can be identified by their behavior compared to the norm, and these can be audiences that can be targeted with personalized pitches.
All this is useful data when constructing the “user experience” in the web site. An online retailer will know the words and layouts needed to tease out the best shopper response, leaving room to present additional information that may be unique to the shopper, based on previous sales or affinity with other items.
A/B testing and web site construction are only now being automated to the point where one no longer needs to write code. “We now have a visual editor. Non-techs can build experiences without developer resources,” Noronha said. The elements of a page can be arranged and text entered without writing a single line of code. All these elements can be tagged to see how users interact with them and that data can be logged and analyzed for insights.
No Diminishing Returns…Yet
For online retailers, Personalization is the beginning of data discovery. Established web vendors like Netflix and Amazon already know where the sweet spots are in their databases.
But for firms starting down this path, they have no idea where they can achieve return on investment and are only beginning to use personalization in their web sites, Noronha noted. They have no traffic numbers to plan by.
Hence the need to find out.