Email isn’t the flashiest tool in the digital marketer’s arsenal, perhaps because it’s been around so long. In fact, it’s been 42 years since Gary Thuerk sent the first email marketing blast (which, incidentally, led to 13 million in sales).
But even though email marketing is officially old enough to be considered Gen X, it remains one of the most popular channels of communication with consumers, even millennials and Gen Z, who overwhelmingly prefer email to other channels.
However, some emails simply don’t land. Here’s what keeps consumers from clicking on your message–and how AI might just be able to help.
They’re too promotional
Everyone likes a coupon, right? According to a recent study by Adobe, the answer is, surprisingly, no. In fact, 39% of consumers said they wished the emails they got from brands were less about promotion and more about education.
And according to Persado, an AI-driven language generator, those numbers are probably spot on, since a recent test of back-to-school subject lines containing messages like “Time’s running out” were some of the lowest performing.
The solution to writing subject lines that resonate with readers without seeming too pushy could be to stop letting humans write them full stop. While that sounds harsh, natural language processing (NLP) and natural language generation (NLG) could be just the tools email marketers need to write and test subject lines that appeal to customers. Some early tests have found machine-generated subject lines to be as much as 98% more effective than human ones. That’s because testing that would take weeks for humans takes just hours for AI, so winning subject lines can go out in a matter of days, a small change that can mean millions in revenue.
They’re not personal
The Adobe study also found that 27% of customers don’t feel their emails are all that relevant. Sure, maybe your emails call customers by name, but so do everyone else’s. While 48% of consumers say that they’re more likely to buy from marketers that leverage their interests, 33% of email marketers have no idea how to do that.
Most email marketers are segmenting their audiences, but breaking customers up into groups and bombarding those groups with the exact same message isn’t actually personalization. New AI-based solutions go beyond typical categories like age or buying behavior to segmentation that inches into the realm of one-to-one marketing, considering factors such as socio-demographics, location, even the weather. The difference is the difference between an email that reads “Dear John, based on the boots in your shopping cart, we assume you like boots” and one that reads “Dear John, it’s going to snow in Denver next Tuesday. Here are some boots in your price range.”
They’re badly written
Ooof. This one’s gonna hurt. No matter how long you spent crafting that perfectly worded email, 23% of your audience says that email was “too wordy” or “poorly written” according to that Adobe survey.
So what’s an email marketer to do? The answer, once again, might be AI. Some machine learning tools can actually predict, in real-time, how likely your message is to get a response based on data from millions of messages. There are all sorts of reasons audiences respond to language in an email. For example, in its back-to-school study, Persado found that used emojis drove higher engagement than emails without, since emojis help make it automatically clear what the email is about.
They’re too frequent
Most of us have to be in the right mood to open an email, and if it seems like a company is too keen, we’re much more likely to ignore a message. 45% of consumers say that being emailed too often by a brand is the “most annoying” thing an email marketer can do.
Even the biggest names in the game can get confused. For example, Amazon recently made for hilarious Twitter fodder when the company assumed a customer who bought one toilet seat wanted a collection of them.
Right now, many email marketers are relying on guesswork for the best times of day to send emails and using outdated automation for product recommendations, essentially crafting emails that suggest purchases they’ve already bought at inopportune times a day, and since we generally receive 121 emails a day, timing (and relevance) might just mean everything.
However, new machine learning tools are making it much easier to predict not only when to send emails but how often by segmenting audiences based on when they’re opening messages and how frequently. The days of toilet seat spamming might just be numbered.
Oh, and if you’re wondering why your emails to co-workers aren’t getting a response, it could be because you included the phrase “Per my last email,” which 13% of respondents found pretty annoying, but…that’s another article.