The more technology advances, the more we expect it to become human. We want our apps and online experiences to know us as well as our friends do.
That’s why we constantly tap the “like” button on Spotify, sending signals that will help it build playlists that “get” us, the way a mixtape once did. When our year-end “Wrapped” summary is released, we delight in an algorithm that knows us inside out.
We look to our Netflix homepage for curated TV and movie recommendations, rather than checking out the latest releases online. And on Facebook and Instagram, we expect our feeds to favor our closest friends and the influencers we like most.
In the world of ecommerce — no less a form of entertainment than our social feeds and media apps — this translates to the products brands suggest to us, both actively and passively, on their websites and via marketing channels.
Advertisement
We’ve become so accustomed to technology that knows us that the experience of being recommended the wrong product is jarring, and frankly insulting, if it comes from a brand we patronize frequently.
But our growing demand for personalization does not exist in a vacuum.
Walking a Fine Line: Where Personalization and Privacy Concerns Meet
Despite the fact that an astonishing 80% of consumers want personalized experiences from retailers, nearly as many (79%) have concerns about data privacy. This puts retailers in a tough spot — damned if you do, and damned if you don’t.
Traditionally, ecommerce personalization vendors function by leveraging detailed user profiles that include Personally Identifiable Information (PII) like a shopper’s name, location, age, gender, as well as more sensitive information like email addresses and credit card details. They connect that data with a user’s online behavioral history to get a full picture of what that person typically shops for and how.
With this info, they can also add a shopper to a segment of, say, women in their 40s in rural England, and recommend items popular with their peers.
However, with the developments over the last few years, from the 2016 Cambridge Analytica scandal to the 2018 General Data Protection Regulation (GDPR) to its North American counterparts the CCPA and PIPEDA, ‘old-school’ personalization based on PII is on its way out — and fast.
For anyone with lingering doubts, Google’s decision earlier this year to withdraw support for third-party cookies on Chrome made it clear that advertisers and businesses must evolve to provide relevant experiences while preserving anonymity.
The Data Economy: How do Privacy Regulations Affect Revenue?
This relatively sudden and sweeping shift in favor of consumer privacy, while warranted, also creates business challenges.
Shoppers don’t just want a personalized experience — 84% of consumers will actually pay more for one. McKinsey reported that “personalization can reduce acquisition costs by as much as 50%, lift revenues by 5%-15%, and increase the efficiency of marketing spend by 10%-30%.”
If brands and retailers can no longer freely personalize their marketing materials, websites and product recommendations based on PII, are they destined to forfeit these business gains?
Absolutely not.
Concerns around data privacy are valid, but they are a signal to adopt less-intrusive alternatives to traditional personalization, not to put an end to personalization altogether.
New Approaches to Personalization
Forward-thinking brands are already experimenting with new ways to personalize while respecting shopper privacy. Here are some of the more innovative approaches:
Gamification
Gamification enables brands to collect non-traditional datasets, like sizing and body type information or personal preferences, in a light-hearted way. This could be in the form of a fit quiz or a “find the perfect fragrance” questionnaire, asking shoppers to answer questions or rate products in exchange for an instant recommendation.
Radical Transparency
Many consumers aren’t necessarily against brands using their data — they just want to know how it’s being used.
The brands embracing transparency as a personalization strategy go way beyond providing GDPR-compliant checkboxes with a clinical explanation of why and how data is being collected. Instead, they showcase how much they value being honest and upfront with customers.
For example, they might explain with a chatbot message that they collect location data to make sure you don’t miss any sales in your area. The goal here is to level with customers and make them feel comfortable enough to share information freely.
Privacy-Friendly Data Sources
When we think about personalization we tend to put the emphasis on the person: their age, name, etc. But what if we could keep the person anonymous and use other data sources to understand what they like?
It’s not about whether Sandra is middle-aged or an urbanite, it’s about her taste, the details that draw her in, the products that prompt her to stop scrolling. Instead of turning to your shoppers to better understand who they are, you can focus on the products they like.
In fact, shoppers themselves say that the most appealing type of personalized content is relevant product suggestions. According to data published by eMarketer, 50% of U.S. consumers want to see products related to their interests; 43% want to see similar products based on purchases and searches; and by comparison, only 17% want to see their name used.
With this in mind, one question remains: What do you need to know about the products your shoppers like in order to make recommendations impactful? It’s all in the product metadata.
The Missing Piece of the Personalization Puzzle: Visual AI
The more detailed your product metadata, the more information you have about the products your customers interact with on-site. That’s where visual AI comes in.
Advanced visual AI technology sees the way a human would, enabling you to scan all the product images in your catalog, identify the items and assign detailed meta-tags based on their minute visual attributes.
Those visual-AI-powered meta-tags can then be used to feed a whole new dataset into your personalization engine. A solution using this technology can provide recommendations that are more accurate than traditional personalization engines by accessing anonymized user data via existing website cookies, and then combining that behavioral data with the deep product data.
Here’s why: Personalization that leans heavily on personal user information recommends items based on SKUs viewed. This new breed of product-focused personalization digs into sub-SKU-level data with minimal use of shopper data, enabling deeper matches.
Instead of recommending just another midi skirt, it’ll know that a shopper likes a trumpet cut, delicate floral prints, silk and a side zip closure.
It’s a revolutionary way of flipping the script to get the details on the items themselves rather than the person, deepening the data without compromising privacy.
Brands adopting visual-AI-powered personalization are seeing a tremendous improvement in business metrics, including higher order values, up to a 4.12X increase in conversion, and 245% higher average revenue per user. With these results, it’s clear that brands can achieve impactful and delightful personalization without pressing shoppers for endless information.
The Future of Personalization for Ecommerce
It may seem counterintuitive at first, but true personalization is not about the people. It’s not about their name or their age or where they live. The human touch comes from an understanding of taste, context and mood, not demographic data.
Spotify recommends songs based not on who I am but on the characteristics of the music I’m listening to, the same way advanced personalization recommends products based on the characteristics of the items I view.
The brands that see clearly into the next era of ecommerce personalization are making a commitment to provide as much value as possible to their shoppers without compromising their privacy. They understand that personalization is not about whether you know their name, but whether you understand what they like even better than they do.
Ofer Fryman is the CEO & Co-founder of Syte. Before venturing into tech entrepreneurship, Fryman worked at Hewlett-Packard, Shuntra, Jungo and Microsoft, bringing with him 22 years of passion in machine learning and deep learning.
