In 2015, technology market researcher Gartner predicted that by 2018, companies that had “fully invested in all types of personalization” would outsell companies that had not by 20 percent, and that by 2020, smart personalization engines that recognize customer intent would enable digital businesses to increase their profits by up to 15 percent.
With the first of those milestones rapidly approaching, I spoke with Paul Roetzer, founder of the Marketing Artificial Intelligence Institute, and John Mihalik, Skyword’s chief technology officer, about their thoughts on content personalization, including how artificial intelligence is already a large part of the marketing tech landscape and how AI can help personalize content at scale yet still provide a one-on-one relationship with individual customers. (Our conversation below has been lightly edited for clarity and length.)
2018 is only weeks away. Have enough companies fully embraced personalization? What’s your take on Gartner’s prediction?
Paul Roetzer: I completely believe that of those who did fully invest, that it would be by 20 percent. But it would be interesting to see the data as to how many we would categorize as truly, fully invested. Standard personalization capabilities have existed over the last five to ten years, mainly in marketing automation, CRM, email marketing, and web personalization based on behaviors and visit history. Most organizations we see have not fully utilized the majority of the personalization tools that exist today. When we get into smart personalization—which I would consider as moving into the realm of artificial intelligence—very few companies are tapping into these resources. I do agree that those who invest in these smart personalization tools would see increased profits, but it’s really hard to find organizations who have truly embraced smart personalization today.
Last year the Marketing Artificial Intelligence Institute came out with your 5Ps framework. What is it, and why did you create it?
PR: In 2015, I attended a SXSW talk where the Associated Press had automated the process of writing earnings reports. I walked out of that session asking myself, can we use AI to create content? Can we use machines to write blog posts?
I aggressively tried to figure out what artificial intelligence is and what role will it play in content production, promotion, and personalization down the road. The easiest definition I have found for AI is from Demis Hassabis, who sold his company DeepMind to Google. He defines artificial intelligence as “the science of making machines smart,” which in turn augments humans knowledge and capabilities. AI is the umbrella terminology for the tools, technology, and processes to make machines smarter. So let’s apply that to a personalization use case for marketers.
If a visitor to your website downloads an e-book, as a marketer you set rules to personalize the experience beyond that. So you say, if the visitor downloads this, then send that visitor these three emails. You have already figured out what goes in those emails, when to send them, and the branching logic of what to do after those emails go out. But you are doing all this work, you are using marketing automation and CRM, you are using what’s available to you today to do personalization the best you can. But the reality is, the human brain isn’t wired to do this. If you think about a single download, okay, it’s easy enough. But if you think about 10,000 downloads of that same piece of content, and consider where those people came from, the different personas you are dealing with, the different stages of the buyer’s journey they may be in, what channels they came in through, have they or have they not been on your website before, are they a customer, prospect, or lead—all these variables go into what content and emails to send each visitor, and that’s where personalization becomes incredibly complicated.
But what we see is that other industries have done this extremely well. Consider Netflix, where 75 percent of what you watch is recommended by an AI-powered algorithm. Netflix even uses AI to generate show ideas and create shows that a machine has told them will likely be successful based on all kinds of behavioral triggers. Or Amazon, which is so deeply personalized that, as consumers, we have developed this level of expectation of how personalized an experience should be that carries over whether we are in a B2B or B2C environment. At the end of the day, it’s human to human and we are used to a certain level of personalization.
AI is literally everywhere in your life, not just in our industry but in our marketing as well. We created the 5Ps framework—planning, production, personalization, promotion, and performance—as a way to categorize the types of AI. When we followed where the money was going and where all the new companies were being developed, the answer was personalization. A lot of early tools are being built around personalization, including what you’re doing at Skyword with your personalization efforts.
Image attribution: Jakob Owens
Let’s talk about Skyword’s personalization technology. John, why did you make the decision to move into this space?
John Mihalik: There’s actually a couple of reasons why Skyword decided to offer personalization technology. The first reason is that consumers have come to expect personalized experiences. In fact, 74 percent of consumers actually get frustrated if the experience is not personalized.
So we have to look at it through the eyes of the consumer. While we as marketers see these types of technologies as ways to maximize pageview consumption or improve engagement, we also have to reconcile the fact that consumers actually expect it now. It’s really a must-have for any sort of large-scale content hub.
In the beginning of personalization, we had ecommerce sites like Amazon, which may have been conceived as a way to maximize profits, but today’s consumers actually see this as a benefit.
However, personalization technologies are difficult to build, and manual configuration is a nightmare. The need is there with marketers to see this same benefit with content, but automated tools are critical to the equation. You simply can’t apply the marketing-automation-based rules engine to the problem of personalization. You would be spending more time creating complex rules than actual marketing and by the time you implemented those rules they would be out of date. What’s needed is a new generation of AI that does what the promise of AI is supposed to bring. It just does it for you. By scanning page and content performance in real-time, AI-driven personalization closes that feedback loop automatically.
Moreover, this same technology can be used not just on websites but also in email campaigns. This is the actual promise of marketing automation. But personalization is not just recommending existing content for consumption. That’s only half of the equation.
If personalization is more than recommending existing content, what else should we be considering?
JM: The other half of the equation is the content you are actually missing in order to personalize the experience. You have to look at what content you are missing. Otherwise you simply have nothing to recommend. True personalization requires AI-infused content creation and AI-driven content consumption. And this fact is almost universally ignored by all other personalization providers.
Personalization requires original content creation in order to succeed. It should identify content gaps for you automatically. We wanted to foster user engagement with our customers’ content as well as provide specific recommendations on what new content should be created. Our recommendation offering was inspired by the needs of our customers to further leverage their investment into original content creation.
What’s your take on the necessary technologies to have as part of a marketing tech stack to build, measure, and personalize content marketing efforts?
PR: If you think about the 5,000 plus companies in the marketing technology space, it can be a bit daunting as to where to start. But this isn’t a situation where you go buy an AI platform that just does all these use cases for you. AI is considered narrow, meaning it’s built to do one thing specifically, like send time optimization, for example. You have to build and train a machine on data sets to do those unique things. You can’t just buy one platform to do it all.
So if you truly want to do artificial intelligence as a marketer, as an organization, you may end up needing to buy a dozen or more AI tools, which is a really complex approach to take. And the reality is that a lot of the best technologies are going to be acquired in the next eighteen months anyway because there is a shortage of AI talent. In this crazy time of AI development, it’s really difficult as a marketer to determine where to spend your time and money.
Image attribution: Matt Barrett
Assuming you’re not in a position to buy a dozen AI tools, what’s your best advice on where to start?
PR: Start with your existing marketing tech stack—go to your automation company, your CRM company, your content marketing platform—and find out if there are other more intelligent tools they offer that would enable you to create better efficiency in your marketing efforts.
JM: You also have to ask yourself, how do these technologies all work together? There has been an explosion in martech. One of the challenges is seeing and ensuring that these technologies integrate together. I’m hopeful for consolidation, but the integration aspect is still one of the most important things to consider. Because it’s not just about how well these things operate on their own, but how they interoperate with your current martech stack.
What are the behaviors and actions being tracked in order to provide a more personalized web experience?
JM: Every behavior is important to track, but conversion behavior is essential. So if someone purchases an item on your website, as an example of a conversion, then the personalization algorithms can back-propogate to determine what led to that act. For example, what content drove that customer to convert and what should be changed in order to drive more conversions in the future.
Lastly, what are the key takeaways you can provide to our readers?
PR: First understand the use case you are solving before implementing personalization and AI technologies. Don’t get overwhelmed. You don’t have to become an expert in AI; you just have to accept that the marketing you do today is going to keep getting smarter and the technology you invest in is going to do the same. Look for opportunities where you can learn more about AI, where you can make your marketing smarter. See this as an opportunity to set a competitive advantage for yourself and your company.
JM: Ongoing content creation is as important to the process as personalizing the content itself. Embrace the fact that AI is a part of everyone’s lives—we need to learn to embrace it, utilize it to help us do our jobs better, and make our lives easier.
For more about Skyword Personalized Recommendations, visit our resources center.
Featured image attribution: Tara Makarenko