Email marketers aren’t meeting expectations for personalization, even after all these years.
90 percent of consumers say they find personalization appealing. And though we’ve made strides in the right direction, we’re not quite delivering the experience they want.
However, advances in several technologies allow us to create more personalized emails than ever before. These advancements are available today, and brands like Coach are taking advantage. Any brand that isn’t, is missing out on delivering the truly customized experience customers want and expect.
Take our quiz to see if your brand is meeting customers' expectations: 5 questions that reveal your email personalization maturity
A personalized email requires…
- Data on the recipient
- Content that fits that data (i.e. is personal)
- A flexible email architecture
1: Even more data
Big Data is old news, but the rate of data collection continues to increase exponentially. Theoretically, the more you know about your customers, the better you are able to personalize content.
In reality, more data alone does not mean more insight. At Epsilon, we believe the email marketing industry has reached a point where better personalization from more data relies on machine learning strategies. To be competitive in the near future, it won’t be enough for a group of marketers to manually identify groups of consumers and create 5-10 “personalized” content pieces for those groups. The demand for personalization will require hundreds or thousands of different versions.
This is where machine learning can help (and has already). United Airlines, for example, uses machine learning to provide engaging experiences to customers throughout their lifecycle. Drawing on historical data paired with real-time data, United Airlines can accurately determine the right communication for every customer.
Combining more data with a machine learning algorithm can identify a customer that should receive certain content, but how can you actually create and deliver thousands of different content variations?
2: Content that fits customer data
Logistics have presented a major challenge with personalization up to this point. To create personalized experiences, marketers have always had to create several versions of content. Moving from tens of versions to thousands of versions compounds that problem. Unless, that is, you can shift away from manually creating every version.
What if you could automate that content selection? By building a library of content pieces that can be combined to create personalized emails, then tagging the content with a matching criteria to your audience, you can do exactly that.
Now the effort of content creation shifts from entire emails to small segments, For example, by creating several variations of hero images that can be combined with text to create a personalized experience.
When a major retailer displays “products you may like” in an email, they’re already using this approach. They have a library of content (in this case product images) that are served to individual recipients based on the recipient’s browsing history, affinity, or other factors.
To truly take advantage of this type of content, however, you’ll need an email architecture that can accommodate these content segments at scale.
3: Scalable and flexible email architecture
Because of limitations in email inboxes, every email template has to work in a number of different email clients. And once the email is sent, content can’t always be changed.
To address these challenges, Epsilon works with clients to create a flexible email architecture. We develop emails to work long term with variable content. Instead of creating a single email for every creative version, these emails provide the bones on which many variations can hang. This approach simultaneously shortens development time and adds the ability to rapidly create multiple versions.
The real power comes when you add the ability to change content in the email after it has been sent. Current technology allows for you to clear that hurdle with open-time content optimization. When your recipient opens the email, they see the content that’s most appropriate at that time, not at the time the email was sent.
A model for the not-so-distant future
Imagine you are a marketer for a national retailer running a large promotion for the weekend. Using a machine learning model, you identify the optimal frequency each customer should receive communications over the weekend: some customers will receive no emails, some will receive several emails a day. You also use a model to determine the items and promotions most likely to drive your customers into the store.
In concert with your creative team, you create content blocks that fit within your flexible email architecture. With your campaign set up, it’s time to hit send. Several deployments occur based on the models you defined earlier.
Once emails arrive in the inbox, you track results. You use point-of-sale (POS) data and online checkout data to see which offers are resonating. Because you can update content in the email that’s already been sent, you feed this real-time data back into your model. The model creates new recommended content.
Content is updated for subsequent emails and in the emails you already sent. Because of your available data, your machine learning engine and your flexible email architecture, you deliver a more personalized experience to all of your customers, without missing potential sales.
Time to get moving
A complex program like this may seem out of reach today, but many brands are already changing the way they email. Email marketers must implement these approaches sooner rather than later. Brands that master personalization using these technologies will build loyalty, while brands that don’t will eventually lose their customer base.