1:1 personalization has become a buzzword in eCommerce today. According to an Econsultancy survey, 94% of businesses believe that personalization is critical to current and future success of the organization.
Experts recommend that online retailers use 1:1 personalization for all three types of on-site shopper journeys—site search, navigation and recommendations. And undoubtedly, when done right, personalization significantly reduces shopper effort, leading to improved engagement, decreased bounce rates and increased revenues.
However, while 1:1 personalization is important and successful, does it work for all shoppers?
1:1 personalization works for less than 25% of shoppers
There are some cases where 1:1 personalization is clearly ineffective, such as in the case of a new shopper, where you do not have enough data to make relevant predictions.
However, in our experience working with some of the world’s largest retailers, we have encountered another large segment of shoppers where 1:1 personalization fails. This user segment is “new to the category.” That is, shoppers who have already bought a product and are now looking for another product in some other category.
The reason for this failure is simple. 1:1 personalization feeds off product and attributes affinity demonstrated by the shopper. But these affinities show only limited portability across categories. For example, consider common attributes like brand and color. It is very common for shoppers to like different brands for different types of clothing; e.g., Docker pants and Polo shirts or black pants and red shirts. While no broad generalization can be made, we have encountered a -100% to +100% correlation of affinities across categories, which means that, in this situation, 1:1 personalization is as accurate as throwing a dice.
For the other 75%, look-alike segmentation performs better than forced personalization
We have seen that online retailers typically have enough data to make relevant 1:1 personalization predictions for less than 25% of shoppers. For the rest of the shoppers, forced 1:1 personalization may backfire as you might predict wrong preferences for the shopper and show them irrelevant products, owing to lack of relevant data.
A natural question that follows is: if 1:1 personalization works in only limited cases, then how do you optimize the experience for the rest of the shoppers?
While you may lack data for personal preferences of these shoppers, they provide many other signals like location, device type, etc., that can be equally or more valuable in these situations. These shoppers should be segmented into groups using these signals and targeted using insights about preferences of each segment. For example, retailers can segment shoppers based on their location and show them the most popular products of the category in that location.
Complement 1:1 personalization with shopper segmentation to provide the most relevant shopper experience
1:1 personalization has immense merit, but only in a few cases. Online retailers must understand both the benefits and limitations of 1:1 personalization to ensure success.
Wherever possible, you should offer 1:1 personalization, and target the rest of the shoppers by segmenting them into groups based on the available attributes. This approach is far more effective than the forced application of 1:1 personalization algorithms to every shopper on site.