The End of Fake Personalization in eCommerce

eCommerce leaders have talked about personalization since I started in this profession – and well – it has basically been BS.

What most of us have been doing is segmentation or some format of auto segmentation - and segmentation sucks.We base everything on incomplete data sets and create shared characteristics from that data; sometimes we may even use our imagination to fill in the blanks. We lump people together in “groups” and present them with products that often are our best guess at what this group wants based on what people like them have wanted in the past.

To make matters worse, we then go ahead and put the label of ‘personalization’ on it... even though it's segmentation at best. Up until now, e-commerce operators have relied on segmentation because predictive analytics have done a good enough job at anticipating behavior... key words in there: “Good Enough.”

When I was at Lucky Brand, we were one of the first customers of a company that became – and is still – the largest player in the world of product recommendations. If a customer was looking for jeans, we suggested more jeans because our logic dictated that customers shopping for jeans sometimes buy multiple pairs at the same time – or multiple washes.

That is fine, but what happens to the customer who is looking to put together an outfit for a 70’s themed birthday party, and needs shoes, a T-shirt, or even a bag? The problem with optimizing solely to propensity to transact or click as your metric of success is you’re not enriching the relationship with your customers. Do you want the person to expand your relationship with your brand, or limit their thinking to one thing? Since when did every single shopper walking into a retail store get presented the same pair of jeans from a sales rep, just because they are all in New York on a Monday… couple of rhetorical questions there.

Generative AI, with its ability to analyze and process large volumes of data on the individual (in real-time), makes a “segment of one” truly possible. AI’s ability to track an individual’s behavior (think: every click and hover) and preferences (think: every page view and every available piece of data about the product) and make hyper-individualized recommendations in real-time is real. With GenAI, I would have been able to understand that a customer looking at jeans is also shopping for a certain look, and I would have been able to put recommendations in front of her that would have mattered to her. More revenue. Conversion rates up. AOV up. LTV up… and we would have created a raving fan that would have amplified the brand with her social circle.

Walking into a luxury store, you get questions, and a collection of products curated to fit your exact needs. That’s what Gen AI can now do, ultimately build a unique product feed for every digital shopper that is adapting in real time. In a profession littered with broken promises – personalization may be one of the biggest. Not anymore!