Scene: It’s 2017, and your CEO calls you in. She asks about your market segmentation—you have five personas, based on demographic data. Then she hands you a report from the data scientist. He’s identified nine distinct segments, based on purchase intent and customer behavior, for which there is an opportunity to increase margin by better targeting service offerings and marketing messages.
You sneak a peek at the methodology, and see statistical and technical gobbledygook:
- unsupervised machine learning
- clustering and classification
- K-NN and K-Means algorithms.
“I thought our data scientist was focused on optimizing our media mix,” you mention. “We’ve improved the performance of our ad spend by 10 percent this year.”
“That’s what you were focused on,” the CEO says. “I asked him about new revenue.”
“Well, this definitely is a great opportunity for the company,” you say, handing back the report. “I’ll get started on this right away!”
She sits back. “Take a moment, and convince me you know enough about data science to lead this charge.” End scene.
Years of experience have caused us to develop an allergy toward selling ideas with fear and hype. However, a few technology advances have seismic impact on the business world and specifically on marketing.
We think machine learning is going to have that kind of impact.
What machine language can do for you
Researcher Tom Mitchell of Carnegie Mellon University famously describes machine learning as the capability of systems “that improve automatically with experience.” The system gets better at its assigned task without requiring any new programming. It learns from the data itself.
Analyst firm MarketsandMarkets says machine learning will be the biggest component in the explosive expansion of the artificial-intelligence market, projected to reach $5.5 billion by 2020, with 50 percent compound annual growth.
Based on our own interviews with experts in academia, at commercial research labs, and in product groups at Silicon Valley giants and startups, we think machine learning’s near-term effect on marketing is greatly underappreciated, in terms of both opportunity and threat. Marketers today mostly think of machine learning in relation to recommendation engines and marketing-mix modeling—if they think of it at all.
It’s much more than that. As in our scene above, it can parse your customer base in ways you haven’t considered. It may be the secret to restarting stalled big-data projects that haven’t shown value. It will begin to automate tasks now considered at least somewhat “creative,” such as writing press releases. Machine learning can also help provide a holistic view of the true efficacy of your marketing efforts, and improve results in specific endeavors like predictive lead scoring—look at Salesforce’s Einstein as an example.
And it will ultimately turn chatbots from today’s FAQ parrots into the actual ambassadors, often the first line of contact between customers and your brand.
These aren’t idle speculations. They’re based on real experiments and developments under way today.
What it might do to you
Of course, riding shotgun with opportunity, there is disruption to the marketing profession.
Every major technology follows a predictable pattern of development. First comes substitution, then augmentation and modification, then redefinition. Think of the evolution of smartphones: First they replaced simpler landline phones, then adapted with the addition of a camera, and ultimately redefined “phone” altogether, replacing cameras, pagers, and PCs for many functions—not an outcome you foresaw when you were toting around your Motorola MicroTAC flip phone.
Expect machine learning to go along the same path. First it will substitute for repetitive, calculation-like activities. In media, there is a lot of mechanical lifting to replace; that’s why programmatic advertising is already machine learning’s first impact crater.
The next phase is augmentation and modification. As media planners and strategists are freed from mechanical tasks, they can focus on understanding how media mix can inform creative work. Right now, all the creative work is done up front; once a campaign launches, it becomes a matter of optimizing placement and timing. Down the road, machine learning may help recognize when the content itself is the problem, and also campaign workflows that are more responsive to news events, for example stopping a programmatic run to lead on-the-fly creative that resonates with a stunt that just went viral at Burning Man or an October surprise in the political world.
The “redefinition” phase is most exciting, and also most dangerous. Machine learning hasn’t hit creative functions yet, but people are certainly experimenting. As it matures, machine learning may redefine creative-marketing work in a way similar to how Photoshop has changed photography.
It’s hard to foresee all the specific applications that will arise in the redefinition phase, but that’s what makes machine learning exciting.
On the flip side, the threat is that your job might not exist in five years—or that you won’t be the right person to do the job, because you don’t understand machine learning’s implications. With the internet in our pockets, buyers’ relationship to products (and the world at large) is different. Some marketers made it through the past decade’s rise of ecommerce, showrooming, search engine optimization, and conversion rate optimization; many did not.
Does this mean that every marketer needs to become a data scientist? No. But it’s critical to understand much more about the underlying ideas so you can use machine learning to your company’s advantage. Nik Rouda, an analyst at the 451 Group, says “it’s a huge mistake to leave it all to the data scientist, who may not understand your business demands.”
There is also an early mover advantage. The faster you apply machine learning in the marketing context, the sooner your systems start learning, and the sooner you start reaping the rewards.
As a small agency, we’re riding this learning curve ourselves. We don’t sell machine learning, and we don’t consider ourselves experts in it. But with our backgrounds in journalism and marketing in the technology industry, we do know how to talk to them.
This column is the first in a series we’ll publish, as we interview machine-learning experts, striving to stay ahead of the curve on understanding how this technology is going to affect everything we do as marketers. We believe that we will benefit from understanding the technology and riding the wave, rather than being dragged along by it.
Next up: a short glossary of machine learning’s most critical terminology, viewed through a marketing lens, to help make your discussions with data scientists more productive and precise.