Ever wonder why after buying shoes online (or any other consumer goods), for the next few weeks or months, you can be sure to spot ads or promotions for those same shoes on nearly every website you visit? What’s more, you’ll see which shoes your Facebook friends bought, which shoes their friends bought and which shoes “others like you” bought.
You already bought shoes, so why are you still being bombarded with ads for them?
The reason is simple: Today’s targeted advertising and website personalization technology tracks and responds to your past actions. This is known as “behavioral targeting.” And it’s the way websites use information collected from an individual’s past browsing behavior to select advertisements to display. Behavioral targeting takes four basic types of parameters into account:
- User Behavior (including past user clicks, searches, page and item views)
- Social Events (such as past user Likes, Shares, mentions and recommendations on social media)
- Item Details (including title, category, price, description and more, of the item being promoted)
- Contextual Information (time of day, weather, device, current location, referral URL or original search query relating to the promotion)
Behavioral targeting then applies this information using one of a number of approaches or algorithms:
- Collaborative Filtering (compares user history — items previously purchased, viewed or liked, etc. — to other similar users and creates a list of recommended items for the user)
- Content-Based Filtering (recommends items with metadata similar to items with which the user has interacted in the past)
The Problems With Behavioral Targeting
The overriding premise behind behavioral targeting is: What people have done in the recent past determines what they will likely do in the future. The problem is that from a psychological perspective, there is little validity to this assumption. Your one-time shoe purchase experience is just one of many examples of how past actions are not the best predictors of future actions, and certainly are not the basis for effective personalization.
But beyond this, today’s advertisers and website stakeholders are asking some tough questions about inherent weaknesses in behavioral targeting, including: With collaborative filtering, how are similarities between users measured? When are two people similar? When they share four, five or 20 tastes?
To what extent do users reject behaviorally targeted recommendations owing to a subconscious assertion of individualism? That is, to customized recommendations based on algorithmically determined similarities, is the user’s knee-jerk psychological response, “I’m not like other people”?
What happens if users delete cookies from their computers or set their browser to block some or all cookies? This can markedly decrease the accuracy of behavioral data and limit the ability to market to individual tastes and preferences.
Today’s advertisers and website stakeholders are asking some tough questions about inherent weaknesses in behavioral targeting.
What if an “individual” isn’t really and individual? Several individuals may use the same computer at different moments.
This is not to say that behavioral targeting can’t provide an effective foundation for personalization. But today, we can leverage advanced user-experience technology to evaluate and micro-monitor on-page visitor behavior. Based on this, we should be personalizing according to each individual’s mindset, affect and level of suggestibility. This will allow us to take personalization to the next level — impacting not only visitor satisfaction and brand identification, but also revenues.
Personalization Based On Mindset And Suggestibility
As I discussed in a previous article, visitors to your website can broadly be divided into two mindsets — browsers and goal-oriented visitors. Which of these categories a given visitor falls into can be measured and determined in real-time based on on-page behavior. Then, content and promotions can be personalized on the following pages he or she visits.
But the fact is that we can drill down even deeper into visitor personality traits, with the aim of even more effectively personalizing content. Even more importantly — we can know our visitors more intimately to make sure we don’t personalize incorrectly.
Only by adding a behavioral perspective to our understanding of website visitors can we effectively personalize content.
A key personality trait that can and should be measured, and which personalization models should take into account in real-time, is suggestibility.
Suggestibility is defined as “the quality of being inclined to accept and act on the suggestions of others.” A suggestible visitor may in mindset be either a browser or a goal-oriented visitor, but once we’ve identified them as suggestible based on landing-page behavior, we can more simply and effectively personalize the following pages.
So, how can we identify a suggestible visitor?
ClickTale has developed a detailed methodology to identify suggestibility, which I won’t reveal completely herein. However, by way of example, on the first page someone visits on your site, a suggestible visitor will first look at what the site has to offer, as a whole. The suggestible visitor will peruse your site navigation and generally show a high level of page engagement. They will hover over special offers, as well as highlighted or promoted content. They will look at “most popular” or trending items, click “play” on promoted page videos and seek complementary content or merchandise.
By way of a more specific example, a visitor to a travel site that first reviews the holiday deals on offer may be considered suggestible. As opposed to a visitor who immediately goes to the flight search engine seeking a specific destination, this visitor is indicating that they are willing to consider what the site is offering, and is open to new ideas.
Once we’ve determined that a user is or is not suggestible, the way we can personalize the subsequent pages they visit is radically different. What’s more — coming back to your shoe purchase experience — the gist of the personalization required may be in direct opposition to your past behavior.
The Bottom Line
To Jarno Koponen’s excellent list of five Personalization Gaps, I believe we need to add a sixth: the behavior gap. Only by adding a behavioral perspective to our understanding of website visitors can we effectively personalize content that matches their mindset, affect and suggestibility. As visitors become more sophisticated and competition continues to grow exponentially, websites that want to survive need to look beyond past behavior to create personalization mechanisms that impact conversions and grow revenues.