Do you want to be a trendsetter or an order taker?
A philosophical question for online retailers
Philosophical question
This is a philosophical question that I have — do you want to be a trendsetter or an order taker?
It’s driven by my observation that many online sites are starting to adopt Artificial Intelligence/Machine Learning (“AI/ML”) to deliver product recommendations and even determine the sortation order on a PLP (product landing page) or a Search Results page.
Online retail is adopting AI/ML
According to Meticulous research, the use of AI/ML within online retail is set to rise 35.9% to reach $15.3 Billion by 2025¹.
Wisdom of the Crowd
For those of you unfamiliar with AI/ML, this technology is often referred to as The Wisdom of the Crowd². In essence, it’s saying that our collective wisdom as a group of people is likely to be more accurate than any one person’s individual judgment.
This kind of approach is often used to create product recommendations that are very common on e-commerce sites. Typically it includes functionality like “people who viewed/purchased this, also viewed/purchased that.”
Issues with the AI/ML approach
Generally, AI/ML-driven recommendations will improve a retailer’s sales. But there are a number of problems with this approach:
1). It’s a retrospective approach. It’s relying on the assumption that past behavior is a good indicator of future behavior. Whilst that might sound logical, it doesn’t apply in all cases. For example, if you’ve just bought a big-ticket item like a new fridge, camera, or exotic holiday, then you are unlikely to be in the market for another one any time soon, although it might be a useful indicator of accessories or follow-on purchases that you could be interested in.
2). It assumes you always buy for yourself. That’s a pretty fundamental assumption, which if wrong could invalidate a lot of the output from AI/ML — i.e. if you’ve been looking to buy a present for an elderly aunt, the recommendations that the AI/ML algorithms are likely to make based on that data are unlikely to be applicable to you when you are buying for yourself.
This problem is compounded when you think about which devices you’re using to browse the web — and what data is being tracked from that usage and assigned to your profile. How many families share the same tablet device in the home to browse online stores and make purchases, in effect aggregating the data of every household user of that device into a single amalgamated profile. If you’ve ever experienced some odd, out-of-context product recommendations, that might be why.
3). Another challenge is the start-up problem, which means that new items added to an e-commerce site don’t have enough data for the AI/ML algorithms to make any predictions. There are solutions to this problem, but they are in effect guessing how something new will perform without having any actual real data.
Biggest Concern
However, my biggest concern goes back to the first problem of the AI/ML approach being retrospective.
AI/ML by its very nature relies on historic data to make predictions about future behavior. It can only see the past, it cannot see what might be a new trend because the data doesn’t exist yet.
It’s a bit like the judges on a talent show trying to make judgments on the contestants (who all basically look and sound the same) against their model of what sells in the music industry — then along comes a young Freddie Mercury. He would obviously be a gloriously successful trendsetter but would almost certainly not get picked by the AI/ML algorithm. You see the problem?
Trendsetter or Order taker?
In reality, if you just use AI/ML for your eCommerce site, then your relative bestsellers are likely to get the most air-time and therefore continue to be your bestsellers, whilst new stock that might become trendsetters given time, won’t get much of a look-in.
Relying solely on AI/ML as your main strategy means that you cease to be a trendsetter and have become an order taker — in effect you’re using the wisdom of the crowd to follow the crowd.
Does it matter?
The counter-argument might be, does that matter if we’re hitting our numbers?
My response to that would be that being a trendsetter bolsters your brand and helps to maintain your price-points and product margin. It also says to the consumer that you are different and worthy of their attention.
If you settle for being an order taker, you’ll be driven by market trends that you can no longer affect. Economics would suggest that in the mid- to long-term, order takers will be competing on price alone, which usually leads to a race to the bottom.
André Brown has a twenty-year background in e-commerce and is the CEO of Advanced Commerce which provides a Merchandising platform (GrapheneHC) designed for Headless Commerce. You can connect to him on LinkedIn or email him at andre@advancedcommerce.io
References
¹ Artificial Intelligence (AI) in Retail Market to Grow at a CAGR of 35.9%
