Master conversion optimization & website testing at Conversion Conference Chicago 2014   Get my full pass for $897!
Home > Content > How Adaptation Can Turn Your Website Into a Money Machine

How Adaptation Can Turn Your Website Into a Money Machine

by Justin Davis, Madera Labs

$2.9 billion.

That’s the amount of money that Amazon raked in during 2006 due to one simple feature on their site: recommendations.

Recommendations are a form of adaptation: delivering content to a user based on that user’s actions instead of simply serving up static content. This kind of thoughtful interaction design can deliver a much more personalized, targeted and engaging experience to the user, not to mention the possibility for hoards of additional revenue from increased conversions.

Obviously, adaptation can be a powerful tool for driving deeper engagement. Amazon’s example is only one of many case studies on the power of delivering content to users based on behaviors, as opposed to a carefully curated content structure. But is it really this easy? Can we really just add recommendations and boost revenue? To answer that, we first need to explore what adaptation is, so we know how it works and how it might be used to increase engagement with our customers.

From Static to Dynamic

At its core, adaptive interaction is the change of an interaction over time, according to some kind of input from the user. While simple in theory, creating effective adaptive interfaces takes careful planning and a thorough knowledge of how adaptation works.

In an adaptive interaction like Amazon’s, there are four moments that act as a kind of “adaptation loop”: Initial interface, user response, system judgment and interaction adaptation. These different moments work together in an endless loop, driving the ability for an interface to adapt to a user.

The loop goes like this: an initial interface is delivered to a user, generally with some kind of content or call to action. The user acts on this (user response), and the system makes a judgment as to the nature of the user’s interaction (system judgment). This judgment is often in the form of evaluating the relevancy of content to a user, based on their actions on it. In short, the system is asking: “Did the user like what I put in front of them?”  The answer to this question is the judgment, which directly drives a change in the interface (interaction adaptation), changing the nature of the interface before the next time the user interacts with it.

Adaptation Loop

The Adaptation Loop

Amazon employs two adaptive modalities: content-based filtering and collaborative filtering. In content-based filtering, the system delivers content to the user based on their consumption of similar content. In this model, the user response is the consumption of other content (any action that defines “consumption” is appropriate: reading, buying, sharing, etc.). The system then makes a judgment based on that action. In the case of a content-based system like Amazon’s recommendations, the user’s action (“purchase”), drives the system to make a judgment: “this person likes X type of book”. After that judgment is made, the system adjusts its output and serves up more relevant content the next time the user interacts with it.

Collaborative filtering is similar, but changes the criteria for user response. In a collaborative filtering system, recommendations are made according to what other users have done, providing a socially-based recommendation system that parallels the kind of real-life interaction a user may have with their friends (“My friend is like me, and they love this book. I’ll probably love it too!”). With this model, the user response is not the active user’s consumption of content, but other users’ consumption of content. This model is more difficult to implement, requiring more advanced models of predictive analysis, but can prove to be a powerful and engaging content discovery method for users.

Knowing this, how can we design more interesting moments into our websites using adaptation? How can we carve out a 35% revenue stake based on the notion that the website should adapt to each user?

A few guidelines:

  • Think about the user’s interaction with you over time. A user’s experience with you doesn’t begin and end with a discreet session on your website. In reality, a user’s full experience using your site has a dimension of time attached to it. When looking at your website, ask yourself how it might change over time to each user.
  • Write a story. When planning your site, use user stories to bring the dimension of time to life. Make up a character and write a narrative describing how they interact with your site over a period of days or weeks. After writing this story, look back and examine it for areas where adaptation makes sense.
  • Start small. Getting this right takes a lot of time. Don’t scrap your current site in favor of a new, adaptive one. Try inserting a sidebar for related content or recommended items. Attach conversion metrics to this to evaluate ongoing effectiveness.

Without a doubt, adaptive interfaces can be very powerful revenue generators. Delivering custom interactions to users based on their behavior creates a hyper-personalized and engaging experience, but it takes time and planning to do correctly. Revisit your website: where can you deliver different content over time to each user? How can you create interesting moments that excite and engage your customers?

Maybe, just maybe, you can carve out your own $2.9 billion stake.

About the Author

Justin DavisJustin has a passion for making things better. From making websites easier to use, to making the grocery shopping experience a better one, he believes that great design is the key to great experiences. With a background that involves technology, music, architecture, and beer (once the host of a podcast about craft beer), this diversity of exposure means Justin approaches problems from unique standpoint, always turning to design to help solve the problem of how to make a customer’s experience better. Justin believes that the customer should be where design starts, and takes a user-centered approach to designing better products and services, using research, observation and interviewing to make sure that the design really works like it should for the consumer. He has a relentless passion for finding opportunities to make experiences better, actively challenging the status quo to help companies move out of their dogmatic safety nets into innovate territories that their customers love.

Justin will be speaking on adaptive content and behavioral targeting at Conversion Conference, March 14th in San Francisco.

Categories: Content Tags:
  1. March 8th, 2011 at 16:13 | #1

    Great Blog Justin, I just wish I could have a small % of your drive, and maybe I can make a buck or two from the Internet. Theres to much overload of information on the internet and like a spider building in there web, you get sucked into some program or programs tha just don’t work. Libertyed :) P.S. Yes I would love a small slice of that 2.9 million in sales….

  2. March 8th, 2011 at 16:35 | #2

    Justin, excellent post! You should visit baynote.com and take a look at our approach. Our Adaptive Recommendations solution marries content-based filtering with collaborative filtering. Baynote understands visitor intent by matching users’ search queries with their website behavior and engagement, segments visitors based on their intent, and recommends content/products that the segment found useful and engaging. You use the exact same terminology and approach that we use :-) Best, Boaz

  3. March 9th, 2011 at 04:10 | #3

    Great Post. Wondering whether Justin can share his presentation in this blog as well

  4. March 9th, 2011 at 11:13 | #4

    Ed – There are definitely a lot of different programs and ideas out there that can suck you in. The key is in finding out which ones have real economic benefit to your company (as in, drive revenues, are feasible to implement, can be managed well, etc.) and focusing on a few. Doing a few things well is always better than doing a lot of things in a mediocre fashion!@Ed

  5. March 9th, 2011 at 11:15 | #5

    Boaz Ronkin : Justin, excellent post! You should visit baynote.com and take a look at our approach. Our Adaptive Recommendations solution marries content-based filtering with collaborative filtering. Baynote understands visitor intent by matching users’ search queries with their website behavior and engagement, segments visitors based on their intent, and recommends content/products that the segment found useful and engaging. You use the exact same terminology and approach that we use Best, Boaz

    Boaz – Thanks for the heads up on your product! Looks great, and I’m happy to hear we’re speaking the same language! Glad to see some people pioneering the adaptive market. Justin

  6. March 9th, 2011 at 11:16 | #6

    Kerala Honeymoon Packages : Great Post. Wondering whether Justin can share his presentation in this blog as well

    @Kerala Honeymoon Packages After the conference, I’ll make my slides available online. You can keep up with that at my blog: http://www.maderalabs.com/blog (look for a post following the conference next week)

  7. andrea
    February 13th, 2012 at 12:24 | #7

    i like the postt!! :D

  1. March 17th, 2011 at 15:35 | #1