Or, Why Only Cognilyze can Recommend New Products Online
Real-world “brick and mortar” stores are always promoting new products. Talk to a salesperson about what you want, and if they can tell you “this just came out, and it is exactly what you want,” they will.
Unfortunately, when it comes to on-line product recommendation, new products are never at the forefront. Stores do promote new products on their homepage and have sections for new products. But as you browse around on-line stores, you constantly see recommendation bars promoting popular products, such as products most often bought by people who share your interests, or products most often bought by people who browsed the same product you are looking at right now. But on-line stores rarely recommend new products in these recommendation bars.
If promoting new products is such a staple of sales in the retail world, why wouldn’t online store want to copy the success, and recommend new products whenever they try to give recommendations?
Consider what actually happens when a salesperson in a physical store recommends a new product that matches your interests. That salesperson has seen the new products received that day or that week, and has understood what is interesting, different or beneficial about each new product. Likewise, he intuitively knows his customers’ wants and needs. This understanding enables the salesperson to recommend the new products to people who would benefit from them.
Online recommendation systems, however, work very differently. While human salespeople think about reasons and interests, virtually all online recommendations are based on statistical correlations. Recommendation labels such as “Customers like you purchased…” or “Customers who browsed this item purchased…” are doing exactly what they say - recommending the products that have been bought the most by customers that are similar to you in some way.
On the surface, this makes sense. If the majority of people that purchased the same things that you bought went on to purchase other specific products, it stands to reason that those things would interest you. Likewise, if the majority of people who browsed the products you’re looking at now ended up buying a few specific other products, it is logical to recommend those products to you.
This is not possible, however, for new products. Simply put, new products by definition have not yet been purchased by a lot of people at that store, so there will not be statistics available to indicate when they should be recommended. Without such statistics, new products can be promoted on a home page or a page for that category of product, but cannot be targeted at individual people, since there are no statistics available to indicate when the new product should be recommended.
For on-line stores, the inability to recommend new products means lost opportunities to sell exactly the products that should be most promoted. For on-line marketplaces, however, such as auction sites and sites that sell products created or sold by individual sellers, the problem is much more serious. New products are the backbone of these sites. On marketplace sites, new products are generally sold and leave the site before they have the chance to build up a base of statistics. For this reason, the ability to recommend new products is important for all stores but particularly critical for marketplace sites.
Cognilyze is the only online product recommendation system that can recommend new products to individual shoppers. This is because Cognilyze is the only recommendation engine that recommends the way human salespeople do, based on reasons and interests and not statistics. While other recommendation engines use statistics, Cognilyze uses psychology.
Cognilyze’s engine recommends new products the same way it recommends all products. First, it analyzes all the products in a store to determine the motivations, preferences and interests that most likely drive each product’s purchase. We call this the “psychology” of each product. Second, takes the collection of products that each user has purchased or otherwise expressed interest, and the psychology of each of those products, and computes a psychological profile of each user. A user’s psychological profile is made up of the motivations, interests and preferences that are inferred from the products that the user purchased or browsed. Third, it finds the products that the user has not yet purchased that match each user’s profile. Since each product is analyzed independently in terms of its underlying psychology, new products can be matched to user profiles just as easily as old products, to find the products that match the psychology of each customer.
With Cognilyze’s recommendation engine, new products can be targeted at customers the first minute that they are available on the site, at the same time that they are being promoted on the homepage and new products tab. And in marketplaces, products added to the listings can recommended to customers as soon as they are added to the site.
Cognilyze’s product recommendation engine was launched with our first client in the summer of 2016, delivering a higher customer conversion rate from the first day it was launched. Mimicking human salespeople, by truly understanding products and customers, beats statistics. The ability to recommend new products in on-line stores and marketplaces is only one of the many advantages of Cognilyze’s psychology-based big data.
