We’re used to experiencing a ‘Personal’ service in interactions we make with physical stores. Yet it’s still something quite rare to find online. The amount of products and services we buy with computers and phones has exploded. Although this convenience has sometimes been at the expense of quality of service.
With personalization you’re optimizing the ‘who’, ‘when’ and ‘where’ of customer experience. Tailoring your product offering to individuals, often grouped by preferences or previous behaviour.
The Dictionary definition of personalization is to “design or tailor to meet an individual's specifications, needs, or preferences”. When used effectively, it gives consumers a smoother and more personal experience online. It offers experiences aligned with user needs and desires.
Another way to understand personalization is to split it into the three forms it can take.
Implicit, Explicit and Hybrid personalization
Do you shop at Amazon, or use a Social Network like YouTube? If so, you will already have experienced personalization. On Amazon, your shopping experience is tailored through your previous browsing or purchasing trends. On Facebook, the content you see is personalized based on friends or pages you interact with most.
These are both examples of Implicit personalization. Implicit personalization is based on what a company has learned about you. This could be page views, purchases or information you've submitted.
You may also experience explicit personalization. This is where a user personalizes the product through features provided by the system. For example; subscribing to a channel on YouTube you personalizes your feed.
Hybrid personalization combines both of these approaches. Facebook’s news feed contains of explicit personalization, as you pick the connections you make. It also features implicit personalization. Facebook decide who’s content to put where in your feed.
Facebook and Amazon might seem like obvious examples of personalization, but if you look around you’ll see it everywhere.
The State of Personalization
When you visit Swoon Editions (an ecommerce site) for the first time, you get a welcome message. It includes a discount for leaving your email address.
This only shows to first time visitors. It encourages them to leave an email address before browsing the site. They’ve personalized ‘when’ to show you the offer.
Another ecommerce site, Working Class Heroes dedicates part of each product page to recommendations. They’re showing different product based on ‘who’ you are. Distinguishing you based on what you’ve browsed or purchased.
Finally, if you’ve browsed a news site recently, it’s likely you’ve encountered a paywall. Below is an example from the New York Times.
They decide ‘when’ to promote subscription, based on how many articles a user has seen.
The New York Times is a good example of a well personalized product. Casual browsers on their first couple of articles aren’t prompted to subscribe. If you’re a heavy reader, the frequency of prompts to subscribe increases.
As you can see, there is a lot of personalization in products we already use. If you start looking for it you will find it in many places. There are still some factors that may put people off personalization.
The Barriers to Personalization
Experian recently published a white paper on personalization. They surveyed 250 technology professionals, to understand the biggest challenges with personalization. They asked interviewees to rank their biggest challenges. The top 3 were; gaining insight quickly enough, having enough data, and inaccurate data.
Personalization is driven by timely insight. A sufficient quantity of clean data is also an important factor, as it’s what you use to segment your users.
Even if your customer base is small, there are plenty of opportunities to collect data and insights. We'll cover techniques for gathering these insights in future blog posts of this series.
The data hurdle is a reason why machine learning for personalization can be problematic. Let's explore the relationship between the two.
Personalization and Machine Learning
When we combine personalziation and machine learning, there can be higher barriers to entry. There is currently a lot of hype around machine learning. In fact, in 2015 it was near the top of the Gartner Hype Cycle
Many people believe that machine learning and artificial intelligence will power personalization. Some even say that machine learning will replace human input in personalization.
Machine learning can be powerful, but needs large data sets to be viable. It’s why Amazon can predict what you’ll be interested in buying. It also means you can have a tough time getting started with machine learning.
Even if you do use machine learning to power your personalization you will still need human input. Humans need to determine hypothesis, and figure out what your goals are. If you’re looking to get started with personalization, there are easier ways. Segmentation is a simplified but powerful way to personalize. It's analagous with A/B testing.
Personalization and A/B Testing
A/B testing is a well established way of optimizing product experiences. Developing a new product feature and then giving it to a part of your customer base. A/B testing only answers the ‘what’ of product optimization.
With personalization you can test the ‘when’, ‘where’ and ‘who’ of product improvements. When it comes to building personalizations they can be tested in a very similar way. First, you create a hypothesis of how to improve your product, you build it, and give it to part of your audience. We’re going to expand on this more in the next post of this series.
How to successfully personalize your product
This post is the first in a series on personalization. The next two parts will be published in coming weeks. The next post will explain step-by-step how to effectively test personalization in your product. The final will address common challenges of personalization, introduced in this article.
Sign up below to get the next part via email.