Just like the front of a physical retail store, the hero image at the top of an online retail store is key to drawing shoppers in to take a look around instead of walking on by. Physical storefronts are meticulously merchandised to attract a broad cross-section of shoppers.
So are online storefronts, except with the benefit of A/B testing, the online retail marketer can test a couple different hero images for any given initiative and optimize to show the best performing.
But now that the shopper is drawn into the online store, how can you provide the equivalent support of a physical store associate? So many different permutations of shopper preferences, products browsed, body language, needs, and concerns. Thinking about manually setting up a comprehensive tree of A/B tests to tackle the millions of possibilities is dizzying.
Enter Machine Learning.
IRL, in the split second before a store associate approaches a shopper to assist them, they may be able to visually observe a couple of things to inform the next words they share:
- Shopper’s body language
- Item the shopper is currently holding
- Top item in the shopper’s tote bag
In that same split second online, Machine Learning is able to observe equivalent signals and so many more in great granularity before choosing the most influential message to share with the shopper:
- Digital body language - How the shopper scrolls, hovers, and clicks
- All products browsed - Including today and previous trips to the store and in what combinations those products are most often purchased together
- Entire Cart - All products currently in or previously added and removed from the cart, in what order, and how long it took to find each
- iPhone user - But using the Chrome browser app which could mean more tech savvy and/or affluent
- Read Reviews - The shopper spent significant time reading reviews vs alternatives like browsing the return policy, which could mean they thrive on social reinforcement
- 100’s of other data-points - Unobservable by the human eye
In about the same time it takes a store associate to make an assessment of a shopper, machine learning can combine and appropriately value hundreds of data-points to discover what would otherwise be unintuitive patterns to the human observer. These patterns are then used to make a prediction of what the shopper needs to see next (if anything) to maximize the likelihood and value of their purchase
That sounds useful, but I also A/B test other parts of my online store like the layout of the Dresses collection page.
Not a problem, just like changing the floor layout of the Dresses section in a physical store, if an associate engages with a shopper, they can take into account that formal-wear has been moved towards the back in favor of prominently displaying the summer fashion.
But unlike in the physical store, online you can use your existing tools to measure both the impact of the A/B layout test AND the incremental impact of machine learning at the same time with proper analytics and attribution.
Additional revenue? Awesome! How do I get started?
You could go back to school for a PhD in Data Science, or hire a team of Data Scientists and Engineers to start building.
Or you can take the fastest path to results and work with an expert partner like Granify to guide you through the best approach to assist your shoppers and drive more revenue.
If you'd like to chat with me more about how to apply machine learning to complement existing A/B testing to assist your online shoppers and drive more purchases, you can drop us a line and we'll set up time to talk!