I’ve been following the increasing attention being given to AI (Artificial Intelligence) and ML (Machine Learning) by digital marketers fervently. If you are a digital marketing professional, and you’re not following it … well you might end up being a Luddite. Sorry, but there’s no time to sugar coat it for you. Things are moving fast.
I must admit that my interest comes with some trepidation. These are game changing technologies. Breakthrough innovations if you will. Every time you have breakthrough innovation, there are bodies left lying around. It is inevitable. Henry ford blew up a lot of good farrier businesses. They were replaced by Goodyear, Firestone, B.F. Goodrich, and others.
I started tracking the Content Marketing space five or six years ago when Joe Pulizzi and Robert Rose began to pick up steam in their quest to redefine it. Pulizzi was a journalist, and a visionary. He saw the need for marketing content, and the discipline of content marketing. Rose is a digital marketing guru. Together they created 1,000’s of jobs for English and Journalism majors, many who were probably still living in their parent’s basements, who could write content, post blogs, define personas, manage editorial calendars, and tell stories.
How will AI and ML effect content marketing. How will machines tell stories?
You can’t argue with the numbers.
I’ve had the pleasure of speaking with Brad Power a couple of times. He is a very sharp guy. I feel like I’m getting sharper just by reading his stuff. The fact that he’s often being published in the Harvard Business Review make me feel smarter yet. One of Brad’s recent posts was How Harley-Davidson Used Artificial Intelligence to Increase Sales Leads by 2,930%. Nope. Not a typo. You can talk about pageviews, Leads to CRM, total conversions, organic site visits, et. al., but some metrics just scream louder.
“Last year, we tripled our business over the previous year.”
Asif Jacobi, Harley-Davidson of NYC
Tripled sales: Yep, that’s a metric that turns heads!
Even if equipped with digital tools and other marketing technologies, humans can only manage a few hundred keywords at a time, and struggle to apply insights across channels with any precision. Conversely, an AI tool can process millions of interactions a minute, manage hundreds of thousands of keywords, and run tests in silica on thousands of messages and creative variations to predict optimal outcomes.
Brad Power, Author & Consultant
Machine Learning takes AI to the next level by training software systems to do the work of analysts. Together, AI and ML then create the conversations that marketers have with their audiences. It is now routine for machines to learn what content is more successful in tests, and automatically serve that content to audiences. Plus, when you layer in the additional complexity of doing this work across multiple social channels, well machines can handle that, too.
Yes, machines are driving what we build our knowledge bases with. Is it just me, or are we flirting with the edges of something that is arguably just a little creepy?
AI and ML algorithms assess and identify the attributes of the best customers, the most active learners, and the prospects that spend the most time on site pages from CRM (Customer Relationship Management) data. Software then aligns those inputs with the data stored about the target audience’s characteristics, and then serves us the content most likely to drive an action or entice deeper engagement.
Here’s a simple example from How Machine Learning Will Be Used For Marketing In 2017:
A hypothetical A/B test can show that after a testing period for a campaign, the audience interacts best when “Welcome” is used instead of “Greetings.” Machine learning algorithms can be applied to the data from all campaigns to deduce the best textual introduction for emails sent to an audience, or even to an individual.
Jeffry Nimeroff, CIO, Zeta Global
But will we know when the author is a machine?
Wow. So, I guess this means the blog post we thought was written by a girl named Beth was actually manufactured by something called Binary? How good are Binary’s prose? Do the paragraphs weave together a cohesive story? Do the nuggets of content that were machine assembled deliver a logical flow? Are the transitions and segues fluid?
The AI software vendor used on this project was Adgorithms, and the software is named Albert. And obviously Adgorithm’s algorithms and Albert’s software are working. Given the results, Albert’s slicing and dicing of chunked copy and graphics from Harley-Davidson’s crack corporate marketing teams have managed to make ‘copy machine copy’ appear to be man-made copy.
Yet, the question about reading a machine’s words warrants further consideration. I’ve got good friends who are copywriters. I enjoy seeing how they weave a story together. If we are being written to by machines does it take the feeling of connection that comes with reading the writing of another human being away from the experience? Will audiences come to recognize copy that is manufactured via algorithms? Will they turn it off as we sometimes do with automated telemarketing?
Case in point: I was driving to dinner with my girlfriend a couple of weeks ago. It was her birthday. Her phone rang. She answered, “Hi this is Carol.”
As we drove for the next thirty seconds she said nothing. Then hung up. I remember thinking how that must have been a great telemarketing intro to keep her on the horn that long. Curious, I asked, “Who was that honey.”
She replied, “My dentist’s office. They called to wish me a happy birthday.”
“That’s really nice of them.” I acknowledged.
She had this annoyed look on her face when she said, “Not Really. It was stupid.”
Like a true glutton for punishment, I dove in, “What do you mean stupid?”
“It was just stupid.” Then she explained, “First of all it was a recording from the receptionist. I like the receptionist, but it was a crappy recording. It sounded like she was in a tunnel. ‘This is Dr. Dentist’s office. We’re just calling to wish three second pause – CAROL – three second pause a very happy birthday. Have a great day pause/CAROL/pause.’ And the ‘CAROL’ was some computer generated voice that was not the receptionist. And the computer voice was way too loud.
Maybe it’s just me, but nothing says, ‘We care deeply’ like a bad auto-dialed birthday wish with a computer generated voice of your name inserted in between long pauses.”
She’s right. We have all had those calls.
In the end though, the data clearly bears out that savvy Albert and his algorithm authors are managing to make it work.
Changes Ahead for Content Marketing
I reached out to author Brad Powers, posing the question, “Do you see this as a harbinger of the end of the content marketer’s job?”
He replied to the contrary, “There is a need for more content to serve more granular and specific offers.”
It looks like there will still be lots of writing – just in smaller and smaller batches. And writers will be tasked with aligning to inputs from A/B tests and machine learning-based keyword analysis. Yet, I still can’t help but wonder if this will compromise the role of the human storyteller. As it appears storytelling is already giving way to writing the building blocks of a story that drives marketing success, versus the design/build of the finished product.
Clearly AI and ML are changing the game. In the case of Adgorithms’ work for Jacobi’s Harley dealership, his audience is likely to be, demographically speaking, very sophisticated. The people buying Harley’s in NYC are able to drop $20,000 or more on a luxury item that can also then cost them thousands a year to store. So the copy coming out of hybrid algorithms developed with machine learning is probably good. At a minimum, unrecognizable as ‘copy machine copy’.
At the end of the day, if the headline “Fat footed ducks ride Harley-Davidsons” drove the right A/B metrics across a broad enough test population, we’d see an ‘Aflac like’ spot with waterfowl wearing Army boots and leathers, driving Hogs down an open road into the sunset, during the Super Bowl. What is working, as proven by the data, is what will primarily be used.
The New Breed Content Marketing Leader?
It is also interesting to see some of the changes being driven by AI and ML showing up in the executive employment marketplace. A recruiter buddy called the other day to learn more about this burgeoning arena of digits, and math, and marketing, and content. He is trying to fill a newly created VP-level position. Apparently, if you want to make several hundred thousand dollars a year plus fat bonuses, be a statistician that can develop strategy and tell stories. The role being filled is at a global advertising and marketing agency. Here’s the job description (note the position’s title):
The VP, Anthrolytics is both a Data Scientist and a Marketing Strategist. You consider yourself a scientist who is well versed in collecting, organizing, and exploring data. You are a strategist who employs acumen, ingenuity, and storytelling to anticipate and answer questions for internal and client partners.
You thrive on data – both big and small — and see patterns that others don’t. You’re curious and want to investigate cause and effect. You’re creative and enjoy finding new ways of doing things. You work in the details but see the big picture. You enjoy explaining things to others and hearing their perspective. You like puzzles and problem-solving.
If you just said to yourself, “Hey, that’s sounds like me!” drop me a line and I’ll facilitate a warm introduction.
I knew I should have taken that statistics class instead of the Wilderness Survival elective in college.