“People don’t exist in a vacuum, and only technology that can think for itself has the ability to adapt to your users’ ever-changing context.“
It was a dark and stormy night. (Actually, it was a relatively mild autumn day in an air-conditioned office.) An email marketer named Michael launched a routine campaign before grabbing lunch with his co-workers. Nothing in particular seemed amiss; nothing ominous seemed afoot.
[Cue creepy piano music.]
When Michael returned, his desk looked just as he left it: his laptop was perched on a stack of papers, and there was a jar of chocolate-covered almonds to his right. He promptly began clearing out his inbox, but he couldn’t shake the feeling that something was off. Was he missing a meeting? He checked his calendar. Everything seemed fine. Was he supposed to call someone about something? Not that he remembered — and anyway, he would’ve set a reminder.
It was only when he checked on his campaign that his stomach dropped:
He’d mistakenly sent messaging meant for Millennial Males to a giant group of Female Baby-Boomers.
[Cue flickering lights. Cue blood-curdling shriek.]
With Great Power Comes Great Responsibility
Marketing automation solutions are powerful, allowing marketers in every industry to scale their efforts exponentially. Our partner Marketo defines it as “a category of technology that allows companies to streamline, automate, and measure marketing tasks and workflows, so they can increase operational efficiency and grow revenue faster.” That’s pretty broad, so I’ll focus on what Marketo calls the engagement marketing engine, which handles “the creation, management, and automation of marketing processes and conversations across online and offline channels.” The key word here is conversations.
As any prolific texter knows, the more conversations you add, the harder they are to manage. And when tasked with managing thousands upon thousands of conversations, marketing automation solutions are capable of causing a lot of destruction.
It’s no wonder so many marketers are dissatisfied with it: according to marketing technology consultant and analyst David M. Raab, a recent survey from Marketo and Ascend2 found that 61% of buyers saw marketing automation as just “somewhat successful,” with 14% deeming it “clearly unsuccessful.” He also cites a Salesforce survey where 31% of participants rated marketing automation as “not very effective or not at all effective.”
Unlike La Chupacabra and those twins from The Shining, marketing automation isn’t inherently evil — far from it. To quote Foundation Capital’s Ashu Garg: “The way I see it, John Wanamaker’s old quote — ‘Half the money I spend on advertising is wasted. The trouble is I don’t know which half’ — is finally both old and out of date.” That’s in large part thanks to marketing technology.
But not all marketing automation is created equal, and some of it is… not particularly smart.
Rules-Based Marketing Automation
The most basic type of automation is rule-based, and the majority of what’s out there in the market still fits neatly in this category.
At a very high level, here’s how rules-based automation works:
- You decide up-front which actions are going to trigger other actions (e.g., “When someone subscribes to the newsletter, send them an email thanking them for signing up and listing that day’s top five articles.”)
- You press go. Yay!
- The program runs, regardless of what’s actually happening. Half the United States could be facing a zombie attack and still the program would run, cheerfully sending emails and alerts. It has no ability to glean insights and learn from them; it simply knows what tasks to accomplish.
Look — rules-based automation has made marketers’ lives way easier, and in a lot of areas, it does the job perfectly. Nothing more is needed. Still, for somewhat more nuanced tasks — suggesting an ideal dinner recipe, recommending discounts — you’ll probably want something cleverer, something that won’t make people feel like they’re being talked at instead of talked with.
Machine Learning-Powered Marketing Automation (New Age Marketing Automation)
The difference between rules-based marketing automation and machine learning-powered marketing automation is the difference between a student who can solve an equation perfectly and a student who can tell when you’re using the wrong equation.
At (once again) a very high level, here’s how machine learning-powered automation platform works:
- You decide up-front what matters most to your business. Do you want people to focus on content from a certain timeframe? Do you want to engage people in a certain geographic region? Do you simply want people to spend more time on your site, or do you want them to complete a specific set of actions there?
- You press go. Yay!
- The program runs, and — by taking in information about the world in which it operates (e.g. your website or app) — it starts to figure out how to reach the goals you’ve set up. For example, if you run marketing for a travel site and your goal is to have people purchase more international flights, the program will start figuring out what it needs to do to make that happen — e.g., it might recommend flights from New York City to the Caribbean if a particular user lives in NYC and shows an interest in warm weather.
- The program gets better and better. “Yay” doesn’t suffice. Huzzah!
Machine learning-powered automation isn’t always perfect. Sometimes, it’ll recommend something slightly off to someone; sometimes, it’ll send a deal that doesn’t make sense to someone else. But these little mistakes are actually key to making it improve over time; it’ll learn from each one. And it won’t make the same mistake for a million people in a matter of seconds.
This kind of automation isn’t necessary for every single task. If you simply want to send new users a welcome message, you probably don’t need to tailor it to each and every person, although I would argue that even at that level it pays to personalize. It’s now a necessity, though, for the modern marketer to educate herself about machine learning and figure out how it fits into her marketing stack.
When it comes to actually getting people engaged with your content and brand, machine learning-powered automation is a no-brainer. People don’t exist in a vacuum, and only technology that can think for itself has the ability to adapt to your users’ ever-changing context.