For years, marketers have personalized their websites. Adding things like dynamic recommendations to their entry and product detail pages, showcasing little slugs that say “Welcome Back, Jack” to the headers, and pushing dynamic upsells based on carted items in the checkout.
Now, with Artificial Intelligence and Machine Learning, instead of little dribs and drabs of personalization throughout your sites, you can customize your entire site from start to finish for every individual who visits.
Can you do this the way most vendors promise it? All at once and overnight without any issues? Hard no. But if you set up a structured system, your dynamically personalized site experiences will be finished – and making money – before you even know it.
What is a structured personalization system?
A structured personalization system is one where you know how the elements work individually and combined. One of the most significant benefits of using artificial intelligence in personalization is that it learns and grows as it analyzes more data. That means your system will be fluid and ever-changing. If you want it to improve over time (instead of, you know, getting progressively worse), you need to set up a solid foundation that measures how each customized element should work and how they all work together.
Doing loads of personalization on individual elements is no longer effective. Was it at one point? Yes, for many companies. And it’s important to remember that today’s AI is lightyears better than it was three years ago. When designing personalization systems, it’s essential to consider how much they will evolve in the next few years.
Which elements are easy to customize with Artificial Intelligence?
Almost every critical element is available for some dynamic personalization. Here are the places you should start with:
Top trafficked pages – home pages, entry pages, and other popular pages on the site. “Popular” often means inquiry and Customer Service pages. Sadly, those are often places that marketers tend to forget.
Carts and checkouts — things like upselling and cross-selling in the cart/checkout and ALL the fields within the cart/checkout. It also includes offering different types of checkouts to different users. For example, Cart X for all Shopping customers vs. Cart Y for all catalog customers.
Navigation — top action bar(s); hamburger menus; bottom navigation; facets, filters, sorts, and refinements; and internal text search.
Pops and Modals — Entrance, exit, and everywhere in between. Please note that pops and modals are “friction,” so if you’ve replaced them on your site, the areas you allocated to do the things you formerly used the pops to do are also ripe for personalization.
Special Offers, Promo Codes, and Coupon Codes, oh my!
Category pages
Product Detail Pages (PDPs) – if you are doing a lot of Paid advertising, you may want to consider doing this first.
My vendor recommended that I do things based solely on the individual elements we’re adding, not on the pages themselves. So, looking at the slammers, we’ve added individually, not our Product Detail and Category pages.
Yes, that’s a typical vendor recommendation. It keeps you focused on how the element is doing and not how it impacts the entire process.
When adding personalization, it’s best to look at the whole picture, not just the frame surrounding the photo. The frame may be gorgeous, but will you still hang the entire thing on your wall if the picture is hideous? Not. So. Much. You need to look at both components, the picture, and the frame.
Also, it’s important to note that although there’s no doubt that personalization works, it can negatively impact website speed/performance. You can also over-personalize specific groups/channels of customers if you’re not careful. This is especially true of Social and Shopping customers.
Plus – and this is the biggie – your personalization elements can conflict with each other, reducing your adoption to cart and thus your revenues. None of those things are insurmountable if you know how the parts work individually and the journey.
How do Artificial Intelligence and Machine Learning make personalization better?
By processing and analyzing copious amounts of data and extracting oodles of valuable insights from the data at scale and in real time, AI/ML has changed the game when creating bespoke sites.
There are many different reasons why AI is great for personalization, but the biggest is that it’s true one-to-one for everyone. Instead of being “the very best thing for most people” showcased to every individual the same way, it’s legit one-to-one personalization. Site visitors get personalized content, forms/checkouts, products, and so on based on their individual past behavior, real-time/current behavior, and what the AI predicts about them.
Best of all, it does this customization seamlessly and at scale. Companies that are great at “regular personalization” typically see 4-6x the results when adding AI/ML to the mix. It incorporates so much more data and continuously improves from its learnings. Additionally, one thing that is not talked about enough is that AI excels at reducing friction. Less resistance equates to streamlined performance, better UX, and an improved customer experience.
One of the other reasons that nobody likes to admit is that AI personalizes based on results (or whatever metric you set for it.) Too often, marketers and merchandisers change things based on their likes and dislikes about specific products. AI doesn’t have those kinds of feelings. It can certainly be influenced by things like inventory, availability, highest margin, etc. Still, it doesn’t say, “even though the purple widget is our #1 bestseller, I’m going to substitute it for _________ because I need to get rid of X” or “because X is one of my team’s products.” AI-enabled personalization is very goal oriented. You set the goal, and it finds the most efficient ways to meet that goal.
The testing is far superior. With AI/ML, A/B split testing is a thing of the past. You’ll be able to test many more variables with greater accuracy and faster than before. You’ll also burn a lot fewer names because AI learns and changes. (If you need a good place to start using AI-enabled testing, email is often the best place.)
It can make and save you money. Because AI-assisted tools tend to be much more focused, they’re also good at reducing DTS (number of days to sale) and increasing AOV (average order value). Both things make you more money faster. AI’s efficiency can save you hard labor costs and reduce the number of contacts/attempts you need to make a sale, thus saving you the associated costs. Because it looks at a customer’s behavior, it can determine which offer(s) are suitable for visitors and which folks likely don’t need incentives.
It provides a better customer experience. I’m pretty much over all this “customer journey” song and dance because I’ve found most marketers come up with these fabulous, comprehensive systems only to implement them in a one-cheeked fashion or, even worse, not at all. The best personalized AI starts with the marketers’ concept, implementing it in real-life and guiding it to improve itself continuously.
As your AI improves, your site becomes more intuitive, engaging, and streamlined. The latter is critical because search engines and other traffic drivers are looking to send their traffic to the sites that are best equipped for it. This loosely translates to sites where the users don’t bounce but stay and add things to their cart or take additional steps (inquiry, quote, download, register, etc.)
Have questions or ideas about using personalization on your site? A tip you’d like to share? Tweet @amyafrica or write info@eightbyeight.com.
Catalogers and other offline marketers have many advantages when it comes to E-Commerce. 
