Implicit vs. Explicit Website Personalization

Categorizing like-minded people into groups then delivering targeted content to each group is called personalization. The first task is to create collections of like-minded individuals and this is the focus of this blog! 

Acquiring Data

Precise data is required to create definitive target groups but how do we collect this? Consider trying to identify if a website visitor is a tinker, tailor soldier, or sailor! Specifically, we want to identify a tinker (Wikipedia identifies these as itinerant tinsmiths). There are options:

1. Simply ask them (explicit) or
2. Collect data from external sources (potentially explicit or implicit) or
3. Observe their website behavior and group them accordingly (implicit)

Explicit Personalization

Explicitly asking a website visitor a few questions when they land on the site is less annoying than it once was. We are already requesting cookie acceptance and research shows over 50% of shoppers say a personalized experience online is important so it is reasonable to expect a few questions would not deter visitors (they can always choose the "Skip" option). This approach will instantly give them the tinker experience they asked for and is by far, the most accurate way to categorize visitors.

We can also pull "characteristics" data from external systems. Classically this would be from the organization's CRM but it could also be from purchased lists or Social Media profiles. CRM data is generally accurate so this is a viable alternative. It also provides the opportunity for the sales team to actively drive targeted content to the website during the sales cycle.

Implicit Personalization

Finally, the observational approach relies on behavior algorithms tracking, scoring, and categorizing visitors into groups (or profiles). A visitor who is a tinker might be calculated as someone who has the following website behavior:

• Selects tinker-related content e.g. clicks the Tinker tab!
• Is interested in remote accommodation (to support their itinerant nature)
• Looked at metallurgy products e.g. a furnace
• Lack of interest in other categories

If a visitor behaves in this way, can we be sure they are a tinker? Are we confident enough to personalize the website to the needs of a tinker (and alienate tailors soldiers, and sailors?). Do content creators need to hedge their bets on what content to deliver? 

Implicit grouping can get it very wrong e.g. a woman begins looking for baby clothing for a friend and is targeted with baby food and diaper promotions. It can also be tricky to transition from one group to another as would be required when a purchase is made, or a visitor's personal circumstances change.

Jaggedness

Implicitly adding individuals to groups can be tricky. This fascinating article at Data Science Central explains the dangers of using averages to make predictions about a given individual. People have so many traits that averaging them to determine their group is likely to result in personalizations that meet the needs of nobody - people are jagged!

The jaggedness principle says that by measuring a trait across a large population, half will be above average, and half below but if multiple traits are measured, across all traits, very few individuals will be average. This is important to keep in mind when developing a personalization strategy. Averages are not sufficiently granular to be used to create like-minded groups used for personalization.

Conclusion

It is hard to beat explicit personalization but the challenge to wide-scale adoption depends on how resistant visitors are to answering personalization questions when they land on a website. Both anonymous and known visitors have complete control of their own experience.

The best alternative to asking questions is driving the website personalization from the CRM. This requires the visitor to already exist in the CRM as a lead or contact and the two systems to be suitably integrated. Typically the personalization data is requested when the visitor submits a lead form. Good integrations allow the sales team to set or override website personalizations from inside the CRM.

Implicit personalization requires relatively complex data science to ensure the individuals are assigned to a group with a high level of confidence. Fortunately, enterprise web content management systems like Sitecore make this easier however a significant effort is still required to ensure the target groups are populated accurately. While implicit personalization works for anonymous visitors, they are usually not able to control their own experiences and it remains to be seen if this lack of control will become unfashionable in the years to come. Perhaps the ultimate solution is a composite of both approaches!

FuseIT specializes in Sitecore to Salesforce integration. Our S4S integration enables both implicit and explicit personalization. Please contact us for more information or to see a demo of this in action.