Optimized personalization is a necessary tenet of any ecommerce brand that aims to stand out. The importance of winning loyal customers depends on one-to-one customer personalization and engagement based on customers’ behavior and intent. As the benefits of personalization become more apparent, the rise in data privacy concerns and their impact on organizations are beginning to take center stage.
Laws and regulations that protect consumer data — such as the EU’s General Data Protection Regulation (GDPR), the removal of third-party cookie support and Apple’s increased privacy protections — are pushing ecommerce operators to find creative new ways to provide successful personalization.
The “Cold Start Problem” asks how personalized experiences can be offered to consumers who are anonymous or first-time users. Since these users make up a significant portion of customers, merchants cannot overlook this demographic. It also asks how “recently added” products can be successfully integrated into recommendation systems when they have no prior engagement.
Reassuring customers about your security systems and ensuring you have all protective measurements in place is vital to the success of your mobile commerce systems as well. One report states that more than 60% of marketers confess to having no preparation to block fraud via mobile marketing, making mobile systems another area that merchants must consider when developing fraud precautions.
Personalization is only as good as the data the merchants can collect. How can retailers optimize the shopping experience with data privacy in mind using anonymous or non-personally identifiable information (PII)?
Personalization with Privacy
It is no secret that merchants must customize their personalization strategies to succeed. Personalization must be fitted to new and anonymous users in our cookie-free future, which is not as out of reach as it seems. On-site behavior creates a rich customer data profile while allowing the merchant to maintain high customer privacy standards.
Using an advanced algorithm, merchants can make intelligent and accurate recommendations based on on-site actions alone. Data remains safe, and retailers can comply with stricter consumer privacy laws. Data that is not PII is the future of personalization and offers the ability to deepen the consumer experience without compromising trust.
So much of this data is more readily available than one would imagine — data is collected by browsers and servers using cookies such as device type, plug-in details, language preferences and time zones. Other sources of anonymous data include personalization according to visitors’ devices, location and referrals.
Of note: Just because a browser can be available globally doesn’t mean that it will be optimized to work sufficiently in all locations. Things such as payment options, security and privacy issues and formatting may need to be reconsidered for different locations in business plans and overall goals. Being creative with what is readily available is the heart of new personalization.
User interactions can be modeled to train recommendation personalization capabilities in the following formats: user-to-item recommendations, item-to-item recommendations and session-based recommendations.
User-to-item recommendations recognize the users’ on-site behavior, such as site search queries, product views, purchases and other behavioral signals. They use that behavior to recommend items to match their needs. Looking at user patterns, items are suggested that are signaled to be relevant. Taking actions purely into account, optimized and personalized recommendations can be provided.
Item-to-item recommendations involve the API fetching similar products from a selected catalog. Product attributes, product text, product images and shoppers’ big data (viewed together, carted together, purchased together) are all included to create personalized recommendations based on items alone. This removes any need for personal data information; the items speak for themselves.
Session-based recommendations employ anonymous user session time data interactions to predict and recommend what might be favorable to the user. From the time they begin interacting with the site, the AI is working to understand these interactions to determine the kind of online user.
These systems involve an implicit feedback model to gather information based on ongoing interactions on a brand’s site. Sources such as clicks, scrolls and page views are the fuel behind this information, and machine learning is based on consumer interaction with the application. This is not data that has to be harvested; this is data that is inevitably being extracted as any individual interacts with the application or website.
Not only does this data not breach any privacy concerns, but it intelligently recycles information already readily available for merchants to use. And when they use this information creatively, personalization is at its most straightforward and accessible.
Zohar Gilad is an experienced technology entrepreneur and executive. In 2013, he co-founded Fast Simon, Inc. (formerly InstantSearch+), leveraging his experience to bring state-of-the-art site search to millions of publishers and e-tailers at an affordable cost. Throughout his career, Gilad has been the driving force behind more than 20 software products for millions of users worldwide.