As an email marketer, you want your messages to show up when subscribers will see them first – at the top of the inbox instead of buried under a mountain of competing messages. Figuring out when to send emails so subscribers will see them is a starting point for email optimization. But if you stop there, you’ll miss a huge opportunity to use this data for greater segmentation and personalization.
That’s what will make your email break through cluttered inboxes and attract attention and engagement.
Not quite. Send-time data uses message opens as a proxy for checking email and suggests general sending times when subscribers are most likely to see them. But this method could mislead you into basing important decisions on faulty data:
If the subscriber doesn’t enable images, the email won’t send the ping, and the open won’t get recorded. You could end up missing a big chunk of data if a sizable number of subscribers seldom or never enable images.
That’s why you need to factor in “moment of open” data. It detects and reports opens based on proprietary elements within the email that get activated when a subscriber opens the message.
Send time tells you when someone opens your email. (See the note above about inaccuracy and under-reporting). Open-time data, integrated with dynamic content, opens the door to deeper segmentation and personalization that can drive the results you want – more clicks that lead to purchases, registrations or whatever action your email should deliver.
You can use open-time data in many ways to drive greater insight and engagement, but these three use cases illustrate how it works:
1. Make email personalization more personal and urgency more urgent.
Adding a first name to the subject line or message copy barely moves the needle these days for engagement. What does? A dynamic content module that changes time-of-day greetings (from “Good morning, Jacinta” for someone who opens an email at 9 a.m. to “Good evening, Jacinta” at 9 p.m.)
Open-time data also helps you get customers moving by using temporal terms (“today,” “tonight,” “tomorrow”) instead of expiration dates, which can be more abstract in the imagination.
2. Test to find the highest engagement time.
Many people check their email first thing in the morning, even before they stumble out of bed. But is that when they’re buying?
Suppose you send a juicy upgrade offer for a mobile phone service. People might see and open your email at 9 a.m. but not be prepared to act on it.
If you set up sending times using open-time data, you might learn your email gets more traction when your recipients are on their lunch hours or in the early evening, when they have time to consider it seriously.
3. Keep customers updated on key developments.
This can be a game-changer this holiday season, especially if expert predictions come true about a surge in online ordering, home delivery or curbside/in-store pickup.
Suppose you send a shipping notice at 3 a.m., based on STO, and your customer opens it at 2 p.m. In the intervening team, the package got delayed at noon. That 3 a.m. email is out of date, but your customer won’t know it based on your email.
Using open-time info, the customer who opens the email at 2 p.m. will see the updated content showing the revised delivery date.
We’re making it easier and more effective to access and tap into this data for your email campaigns and journeys with our new tool, Insights. Want to learn more? Request a personalized demo!
Machine learning is one of the most talked about “new” strategies and technologies in the industry today, but we often find that it is highly misunderstood — and many times, that’s our fault on the vendor side.
It’s our goal to clear up machine learning and decipher what it means for marketers and the amazing benefits it can provide on even a day-to-day basis.
Machine learning is simply a way to apply an algorithm on top of data to process and make decisions faster than what a single human can possibly accomplish.
Besides the misunderstanding of what it actually is, what keeps marketers from leveraging machine learning in email marketing? Here’s a few of the answers we hear across the country:
If you’re experiencing any of these problems as well, you’re not alone!
One of the leading issues we hear marketers say about leveraging machine learning for email is that they have a data problem. No matter the size of the organization, small, medium or large, there’s a “data problem” in all of our businesses! We like to dig a little deeper and ask what the problem is. And we consistently hear there’s “bad” data, too much data, or not enough data to get started.
Everyone — even Amazon — struggles with their data:
She’s a mom blogger, and a mother of 2 young boys. Apparently, she bought a toilet seat. And after completing her purchase, she started to get messages about buying more toilet seats or toilet seats that she may like! She bought a toilet seat out of necessity, not as a collector. So getting these types of messages from Amazon is not only annoying, but also just plain stupid.
Others commented about their own woes with Amazon, getting messages about more burial urns, coffins, or even punch-down tools.
But it doesn’t have to be this way.
You can utilize machine learning for email without causing even more data woes, costing your budget an arm and a leg, or having to deal with infinitely complex tech stack integrations.
So where do you start?
It helps to think with the end in mind.
Machine learning makes the complex, simple. The above illustration shows all the factors and components that go into applying machine learning to marketing.
But what’s important about this illustration is not all the factors and components, rather it’s the fact that machine learning does all the work for you. Machine learning helps ingest, process, and make sense of all your data. Whereas before you would spend time pulling and organizing raw data and then uploading to your marketing technology, machine learning completely eliminates that step for you, so you can focus on what’s most important.
This is the model we use at Cordial to help illustrate all the work that goes on behind the scenes that translate into actionable tactics you can put into place to become an email all-star.
Knowing what’s possible, here’s three ways you can use machine learning technology in real-world, applicable ways for your business.
Testing and optimization is one of those topics that everyone talks about, but no one actually does.
Why don’t marketers test and optimize messages more frequently? It can be time-consuming, inconclusive, and or near impossible with the technology you use today.
Testing messages doesn’t usually produce enough revenue results to justify the amount of effort that is put into creating the new content and campaigns. It’s all about results at the end of the day for each one of us. So, we thought about this process that marketers go through when initially rolling out machine learning to optimize messages to yield statistical significance — or in other words — results!
At Cordial, we take testing and optimization seriously. Which is why we worked to make it as easy as possible to leverage machine learning across subject lines, content, hero images, and promotional offers to continuously optimize towards the desired goal in your campaigns. The machine learning algorithm utilizes a method of testing called Thompson Sampling, which is based on an algorithm called Multi-Armed Bandit Theory.
But you don’t need to remember any of that — all you need to know is that the algorithm works to automatically optimize a message based on the best performing combination of factors.
Using machine learning for testing and optimization also directly applies to triggered and automated messages. You may have triggered automations like your welcome series, abandoned cart, or upsell series that haven’t been touched in months.
Because of the automated nature of these kind of messages, they make for the perfect message to use the testing and optimization machine learning technology above. Since these messages constantly go out, it’s paramount that these messages are optimized to perform as best as they can.
Not optimizing triggered automations is almost like leaving money on the table. Thanks to the application of machine learning to programmatically test and find the best performing variation of a message, you can easily test and optimize your triggered email automations to ensure they’re performing as best as they possibly can.
Finally, product recommendations are one of the best ways to utilize machine learning. Product recommendations are a magical feature that can be a game-changer for your personalization efforts.
Cordial Recommendations uses machine-learning technology and predictive modeling to deliver product suggestions based on real-time customer interactions and behavior. Recommendations improve and adapt over time to increase opens, clicks, and purchases as more is learned about your customers.
Related: Machine learning is just one of several game-changing opportunities for email marketers. Read more on Four Steps To Becoming An Email All-Star.
REVOLVE migrated to Cordial after using a legacy ESP for years and uses the platform to send promotional and triggered messages to their customer base. Because of the gaps in their legacy tech they were unable to provide their customers with personalized or triggered messages, severely limiting their marketing efforts. It simply was impossible to scale personalization using the segmentation model they followed previously
Cordial enabled REVOLVE to completely rethink how they communicate with their customer base, relying heavily on triggered communications that engage customers in the moment they are primed for action. They now use programmatic templates to personalize millions of messages each day.
Since signing with Cordial, Revolve has pushed live 15 different personalized triggered campaigns which now account for over 20% of their overall revenue. These campaigns have a 2X engagement rate over traditional promotional campaigns and have resulted in massive gains in efficiency for the team. Instead of building promotional campaigns one by one, they can rely on Cordial to automate and deliver messages for them, freeing up time for the team to be more strategic and effective.
They also leverage Cordial Experiments to optimize all the triggered automations they implemented and can now be confident that they’re generating as much revenue as possible. It all wouldn’t have been possible without the use of machine learning, which has streamlined their efforts to optimize and personalize messages in a repeatable, scalable way.
To become an email all-star, it’s time to move past the old way of doing things and adopt the new, namely, machine learning technology. Think differently about testing and optimization using machine learning to finally move beyond A/B testing. Leverage your data to learn what your customers want to buy next. Machine learning doesn’t have to be scary, confusing, or out-of-reach.
Let the machines do the work!
Allison serves as Cordial’s founding member of the partnerships channel and growth ecosystem. She entered into the world of email in 2013, after spending time being a creative designer for print media, hopping into technology sales, and being an entrepreneur of a local fitness club. Allison works closely with the leading technology in various industries, giving her an unparalleled perspective in the space.
Cordial is a next-gen messaging platform that helps marketers leverage their data to create timely, personalized experiences for their customers across channels. Instead of relying on multiple technologies and messaging providers, Cordial enables brands to simplify their processes by consolidating promotional, triggered, transactional, and lifecycle messaging to create unified brand experiences that make the customer the center of every interaction.
You can find more content like this on the Cordial blog.