2015 was the year that privacy went bust. The collection and use of personal information has hit new highs, or maybe new lows. The flights we search for one minute are showing up in ads alongside our e-mail a minute later. Products referred to in e-mail messages or Facebook updates show up in ads on newspaper sites minutes later.
At the same time, however, none of this invasive technology gets any closer to knowing what we really want. I bought a heavy black raincoat at one online store in November, and within a week I got e-mail from that store with their “just for you!” recommendations, suggesting I buy one of three other black raincoats. In planning for a business trip to a colder climate, I really could have used gloves. When I bought a new smartphone, I was shown ads for the most popular accessories, but not the accessories that I would ever want to buy.
http://modernhomesleamington.co.uk/component/k2/itemlist/user/31772?format=feed Orario di partenza: entro le ore 11.00.
Bottom line, all the personal data being collected is not enabling web sites to really know what we want.
But 2016 is poised to be a watershed year in customer understanding. Cognilyze is bringing to market its psychology-based product recommendation engine for e-commerce sites. Instead of taking the same approach that today’s recommendation engines are taking, Cognilyze is taking an individualistic and personalized approach to understanding what individual customers want.
Existing systems all use variants on a technology called collaborative filtering, which dates back to the early 90’s. The modern form of item-to-item collaborative filtering basically says to recommend you the products that were bought most by the people that bought the same products you bought. This makes sense, and it works to a certain degree, but unfortunately the products most commonly bought by customers who bought coats are other coats.
Cognilyze’s technology takes another approach. If someone bought a heavy coat, they may be traveling on a business trip to somewhere cold, or they may be going skiing, or they may be replacing a coat to wear around home during the winter. If they buy a red, coat, they may be attention-seeking by nature, or red may be “in” in their social circles. If they buy a coat without a hood, they may be conservative by nature, or they may be buying a coat to wear to work and be worried about maintaining a professional image. These and other details together paint a much different picture of each customer, based on conclusions about why they bought the products they bought..
Once each customer is better understood, they can also be recommended products that match their motivations and preferences much more closely. Someone going on a business trip to a cold place may want professional-looking warm gloves, while someone going skiing will want ski gloves. Someone who is attention seeking will want certain things, while someone worried about maintaining a professional image will want other things. Someone who is extroverted will likely want the most outgoing products in any category that they search or browse, not only for winter clothing. Someone who is trendy will likely want the latest hot brand of shoes on the market, while someone who is conservative will want established brands.
Bottom line, the psychology that underlies customer purchases, the “why,” makes all the difference in what recommendations will catch their interest. A system that can determine customer motivations and preferences will generate recommendations that meet customer interests and thereby make more sales.
There are other advantages to psychology-based recommendations. For years e-commerce analysts have been discussing the “long tail” of less popular products. The short head of popular products are stocked in stores, while the long tail of less popular products are more easily sold online, where shelf space is not an issue. But recommendation engines using collaborative filtering, based as they are on product popularity within clusters, will never recommend products in the long tail. Cognilyze’s individualistic approach is agnostic to popularity.
Cognilyze’s psychology-based recommendation engine has achieved four times the click-through rate than recommendation engines current in use, in trials with major retailers. We are currently working with more retailers to get the system into broader use.
If we’re going to be receiving ads, e-mail promotions, mobile notifications, and all the other forms of product recommendations, wouldn’t it be great if the recommended products were actually the things we want?
With technology like Cognilyze’s, 2016 can be the year that e-commerce sites truly begin to understand their customers.