Think about the last time you researched something you were looking to buy online. How many different sites did you visit? How many product pages did you browse? How long did it take from the beginning of your search to the end?
Now think about all the data that was generated by your activity. Ideally, an innovative online retailer would collect and interpret all of this data, figure out what you’re trying to do, and present you with the most relevant experience to help you accomplish your goal.
Expand this to all the data generated by all shoppers and all of their activity. Shoppers rarely come out and say what they’re looking for, so there’s a lot of noise in that data. Some shoppers start their search by browsing one product or category but ultimately switch to another. Others may be interested in one product but aren’t ready to buy until they purchase something else first.
Machine learning can help retailers make sense of this data and leverage it effectively to personalize each shopper’s experience. But to be effective, you need to give your machine-learning algorithms the right input. What each shopper has clicked on is important, but engagement provides much richer data for machine-learning-driven personalization. Let’s go through an example.
Clicks vs. Engagement Example
Let’s say that I’m considering some upgrades in my kitchen, and I’ve been spending a lot of time on a home improvement site. If the retailer could only track my clicks on the site, this is the story the retailer will see from my actions:
Looking at the wheel in the image above, you can see that while I’ve been browsing on the site I’ve looked at more stainless steel products. I don’t appear to have any real brand preference, as I’ve looked at Samsung, Kohler and GE Profile products almost equally. And I appear to be looking at refrigerators, dishwashers and faucets in my research process, though I seem most interested in refrigerators. I also appear to be looking at both side-by-side fridges and french door fridges equally.
Now imagine that the retailer could understand my browsing behavior based on how much active time I spent looking at each product, reading reviews, inspecting product images, scrolling up/down pages, and moving my mouse. My profile would tell a very different story:
The same wheel now shows that I’m substantially more interested in side-by-side refrigerators than any other product, and specifically the GE Profile brand (a high-end brand on this home improvement site). The other categories may be relevant to me, but are clearly further down my list. For example, I may be looking to replace my refrigerator immediately, but am considering a full kitchen remodel in the future. It is essential to understand this entire context and create relevant experiences that cater to me as an individual.
As you can see, looking at time spent engaging with product content allows you to determine the categories, brands and products that I am most interested in, and not be distracted by the noise of all the clicks that I’ve generated on the site. If this retailer were to rely on clicks alone, I’d likely be shown both Samsung and GE Profile equally, as well as both side-by-side and french door refrigerators. In this case, to put it in perspective, it’s the difference between showing a $1,500 vs. $8,000 refrigerator!
How to use time spent to create a unique and relevant experience
So it’s clear that “active time spent” gives us more information than clicks alone, but what can the retailer do with this information? Here are some examples that can be leveraged across the site by employing machine-learning algorithms:
- When I first land on the homepage, the categories that are highlighted should be based on the categories I have shown the strongest interest in based on where I’ve spent my time. The homepage also provides a canvas for introducing other categories I may be interested in based on my behavior and that of other shoppers who show similar traits. So, for example, the retailer can recognize my kitchen remodeling intent and recommend categories based on what has engaged and converted other kitchen remodelers.
- Brands featured on the homepage should include the higher-end brand I was looking at as well as other brands within a similar price range, because it would be a missed opportunity to send me down a path to cheaper options if I’ve demonstrated interest in something more expensive.
- Content recommendations across the site can feature refrigerator buying guides, kitchen remodeling tips, etc., leveraging machine-learning algorithms to match content with my interests (and that is correlated to conversions).
- The main navigation of the site could feature “refrigerators” prominently under the “appliances” section.
- When I begin to type into the search bar, products displayed within the search bar and search results page should be sorted based on my affinities. For instance, when I search for “refrigerators,” stainless steel, side-by-side, and high-end brand refrigerators should be listed at the top versus showing refrigerators based on general site-wide popularity.
Machine-learning algorithms have the ability to completely personalize site experiences for your shoppers. But those personalized experiences will only be as good as the data you collect. Clicks are fine but they may tell the wrong story. Engagement data based on active time spent is what’s most critical to truly understand each visitor’s affinities and intent on your site.
To learn how Evergage can help you uncover each shopper’s affinities and intent on your site and help you create individually personalized e-commerce experiences, request a demo today.