Products and services personalize to create highly relevant experiences with users in order to provide better engagement, experience, retention and ultimately growth.
Personalization is not analogous to customization and configuration, but personalization along with customization is useful in providing relevant experiences to users. Customization and configuration can be thought of as explicit feedback. Implicit feedback plays an even more important role in making a product or service personalized for users.
In a broad sense, personalization is the ability to individualize experiences which includes content, messages, user interfaces, products, etc. based on both explicit and implicit feedback.
- Examples of explicit feedback are giving a 👍🏼 or 👎🏼 rating to a song, or choosing a certain layout and font size in your favorite editor.
- Examples of implicit feedback are clicking a call-to-action button, observing watching habits, observing browsing activity, scrolling past an article in a feed faster or slower than average, spending 50% more time on an article than average, liking a product, using a product at night between 8 and 9 pm, etc.
Personalization on a daily basis
Personalization is ubiquitous and we see varying degrees of personalization on a daily basis.
Personalization is one-on-one marketing at a massive scale
Personalization is one-on-one marketing at a massive scale. Some are subtle, useful and relevant and others are outright creepy and untrustworthy. Relevant personalizations are a win-win for both marketers and users. Irrelevant personalizations do more harm than no personalization at all.
The age old and simplest, addressing users with there first name – “Hi John, Would you have the usual latte?” – works great at a Starbucks, but may not impress a mobile app user or add value if you are reaching the user directly on their mobile device via push notification or SMS.
Although, addressing a user with a name is personalization, we are focused on more meaningful personalizations in this blog. Amazon for example, makes the communications messages meaningful. “Your package with Philips Sonicare and 3 more items, will be delivered tomorrow by 8PM.”
Redirecting novice users to a wizard like interface and experts to a data-entry page
“You have not used the website in a while. Here is a simple way to enter your expenses. If you’d rather use our expert UI click here”
Personalizing best time to reach and channel
Reaching out to users at the time they would be most responsive based on past behavioral data. E.g. a user may respond to texts in mornings and another user may respond to in-app promotions on weekends. Choosing a different communications channel based on desired outcome, reachability and responsiveness. E.g. email channel for some users, push notifications for others users, and SMS for yet another group of users.
Responsive Content Personalization
Serving responsive content, differently to web visitors and mobile web users. Formatting for small screens, not relying on hover tooltips, finger friendly buttons etc. are common examples.
Relevant Content Personalization
Showing different landing pages to users based on search keywords and what you know about a user, e.g. newness (is it a new user or if returning how long ago did they come before), location, device etc.
Personalization based on purchase history
Personalizing content based on purchase history is a known way to market relevant accessories to users. Offering Nikon lenses to a user who has previously purchased a Nikon Camera body makes sense.
Personalization based on user traits
Serving content based on user traits like demographics, device technology (Technographics), usage and behavior, psychographics (lifestyle, attitudes) etc.
Personalization based on Sensors & Artificial Intelligence
Serving content based on geolocation and geofences.
Changing the app behavior depending on context derived from geolocation, time, and motion sensors. “Here is a mix we created just for you, for your morning cardio at 24 hour fitness”. The app uses phone motion sensors, time, geo location, database of geo fences of gyms, app genres to determine a songs list. Note, some users may find this creepy, ask for permissions with a valid explanation and have a fallback when the user does not provide a certain permission)
Changing the app behavior depending on proximity to external sensors, like an iBeacon “Seems like you at Sushi-o-sushi, would you like to search for promotions”
Changing the behavior of an app when it is connected to a car or a home stereo dock
“This is the first time you have connected your phone. Download BMW Apps from the app store.” Changing the app behavior when you connect to the home wifi? “The kitchen lights have been turned on”
Changing the behavior of an e-commerce app when in-store versus at home. An app should be able to detect the environment where it’s being accessed and offer users appropriate in-store or online context-aware experiences automatically.
Displaying relevant points-of-interest on a map based on past searches learning and predictions. If Mark frequently searches for sushi resturants, the map could show sushi restaurants near Mark when he opens the app during lunch time. Similarly Mary could be shown a jogging path and Mike the time it would take him to pick up his kids from school.
Personalization versus Configuration
Configuring a car is not really personalization, but it gives clear insights to the marketers what an individual or groups of individuals are selecting.
Serving content based on explicitly stated preferences is using explicit feedback to drive personalization. Apple Music, LinkedIn Pulse have long uses explicit feedback to jump start building personalized content for new users.
Here is a summary of a few of the techniques used to personalize.
- Behavioral targeting of individuals – Personalization techniques include extracting tags from content a user consumes and serving new content that contains the same tags. Extracting tags could be discovered explicitly (asking users) or implicitly discovered. Songs, Stories, Product types, Blogs use this method.
E.g. Tumblr uses this method
- Recommender Systems – Recommender Systems aim to improve customer experience through personalized recommendations based on prior implicit and explicit feedback. Users are first classified into groups and then the personalization is based on the group they belong to.
E.g. Amazon uses this method for “people who bought this also bought…”
- Collaborative filtering – Collaborative filtering is predicting a whether a user would like a product based on products liked by of similar people. Collaborative filtering however does not work well with too much (scale) and too little data (sparseness). The sparseness and scalability issues have been solved by the alternating-least-squares with weighted-λ-regularization (ALS-WR) algorithm Large-scale Parallel collaborative filtering for the Netflix Prize. Collaborative filtering can go wrong and produce creepy results.
E.g. Netflix uses CineMatch and a variation of Collaborative Filtering.
Pinterest uses a combination of behavioral targeting and collaborative filtering to serve new pins in realtime based on user actions within the same session.
- Curation – After classifying users into groups based on segmentation techniques for example as outlined in my article, curated content is shown to users.
- Algorithmic curation – When curation is accomplished by an algorithm using traditional supervised machine learning.
- In-session Personalization – Personalizing content when a user scrolls down a web page or moves to another page is becoming a popular way to keep providing value to users. I’ll blog about this in the future.
In-session personalization is known to drive engagement
In conclusion, Personalization is essential for making a service or product relevant to individual users. It is known to increase conversion, engagement and retention – the three critical ingredients of growth. It is never a good idea to over personalize. You have to create a balance between security, privacy and personalization. Avoid invading customer privacy by personalizing on sensitive topics like pregnancy. Personalization should always help the user and marketers should only collect information they absolutely need and be transparent about the information collected.