5 Practical (and Unpretentious) Ways AI Will Change Marketing by 2020 | Hootsuite Blog

5 Practical (and Unpretentious) Ways AI Will Change Marketing by 2020

Blog   /   Social

“I’ve read all your lyrics.”

“You’ve read all my lyrics.”

“My analysis shows that your major themes are that time passes and love fades.”

“That sounds about right.”

“I have never known love.”

“Maybe we should write a song together.”

It’s clear: the marketing machines are here.

From IBM’s Watson chatting with Bob Dylan (above) to robot reporters churning out financial and sports articles, AI is rewiring how marketers plan and execute campaigns.

But should your organization really invest in AI? Or is this only a play for brands like IBM, Google, Facebook, Spotify, and Amazon?

I wanted to learn more about the viable use cases of AI—applications that can bring organizations value today, not in five years time.

In this article, I share the five use cases that are gaining traction across industries. These are based on Skype interviews with leading AI companies, content marketing experts, and analyst reports.

1. Robot reports  

Hootsuite creates a lot of content. And we spend a lot of time measuring how much traffic it drives, how many social shares we get, and how much revenue it creates.

AI could obviously help to analyze our content performance. But we’re not a consumer brand like Google, Spotify, or Amazon. So I was skeptical about whether AI could deliver practical reporting uses for an organization of our size.

Quill changed my mind. In a few seconds, this machine spit out a Google Analytics report. It was well-written, dug up some mobile data I hadn’t even considered, and revealed what caused a revenue lift on a recent advertising campaign.

I’ve worked in agencies and a large part of your time is spent digging through analytics, presenting data, and summarizing what campaigns are driving traffic. Quill delivered a comparable report in a few minutes.

I’ve since set up Quill for our team. This saves us a few hours a week. Not bad for a first introduction to using AI to automate tasks. If you have a Google Analytics account, test it out free here.

2. Automating enterprise content  

By 2018, Gartner predicts that 20 percent of all business content will be authored by machines.

I wanted to find out more about Quill, so I reached out to Narrative Science, the company that makes the solution. Narrative Science helps enterprises  use natural language generation to analyze and tell stories from their data.

“We hear all the time that content is king—but what’s really helping many of our customers is that AI can help automate a lot of high-quality content at scale,” Katy De Leon, Narrative Science’s VP of marketing, told me via a Skype interview.

“This saves time for employees. But also delivers a better customer experience as content is accurate and more relevant to their stage in the customer journey.”

For example, a large organization might have thousands of products. It’s always a struggle to make sure product descriptions are up-to-date and optimized for search and mobile. “Automation technologies can help your team write these descriptions and manage of all this content,” says De Leon.

“When you put out one piece of communication like a blog post or press release that is read by a million people, that is not the best place for AI,” she says. “But when you put out millions of pieces of communication that need to be personalized for each person, that is where AI companies like us shine.”

A popular use case comes from Narrative Science’s financial services customers. The platform can create individualized investor reports for millions of clients, telling them how their bidding has gone historically and offering specific advice for improving their performance. Advisors can also use these reports to prep for calls with clients.

3. Finding patterns in your data

“AI is really good at automating those manual tasks people hate to do. But AI is more than an efficiency play,” Narrative Science’s chief scientist Kris Hammond told me.

“Because AI can analyze datasets in seconds, the machines are providing insights that are simply not humanly possible to find in data,” Kris says.

For example, AI platforms can look at millions of transactions and predict a customer who is about to churn. Or they can analyze an enterprise sales cycle and predict which deals you’ll close in the middle of a sales cycle by taking a look at historical conditions.

“Data is a perfect match for the machines,” says Kris. “You already have talented human analysts in your organization. But they have higher value activities to do than manually look through rows and rows of customer data. The machine can work faster, understanding customer trends instantly.”

I asked Kris what organizations can do to get started with AI. He offered three steps.

  1. Automate the things your team hates to do. These are the tedious tasks at the bottom of your skillset. The machines can help here.
  2. Don’t automate your job. Automate what you need to report on. Most people like their job but don’t enjoy constantly creating reports about what they do and the value they create. You can free up resources, creating more value instead of reporting on what you did last week.
  3. Look at the task and ask: am I making decisions that are mechanical? If the answer is yes, then that task is a good candidate for AI.

4. Reducing duplicate work across the enterprise

As AI is rewiring the way enterprises create and analyze content, I wanted to know about the impact of all of this on SEO. This led me to MarketMuse. Their company uses AI and machine learning to help enterprises refine their content planning process.

Aki Balogh, co-founder and chief product officer at MarketMuse, explained that AI is making the old tools of SEO obsolete.

“In the past, you’d have an SEO specialist team, a digital strategy team, and a team of copywriters,” he says . “Typically, these teams work in silos. For example, the SEO technician delivers a spreadsheet of keywords and then the writers work on ranking for different high-traffic terms.”

As Balogh explained, Google’s algorithms have been working towards understanding concepts, not ranking individual keywords. This is why enterprises need a holistic and integrated approach to researching and planning content marketing campaigns.

“Human brains tend to solve problems by using association and heuristics, applying knowledge and biases from past experiences,” he says. “But AI is mechanical: it can move through the data and spot patterns that we either ignore because of biases or that would take months of manual analysis to spot.”

For example, perhaps your SEO specialist ignores certain keywords as they have a low traffic volume. Or your writers avoid certain topics as they don’t have experience or interest in those subjects.

“Our customers use this intelligence to craft more impactful enterprise content plans,” Jeff Coyle, MarketMuse’s co-founder and chief revenue officer, told me.

As Coyle explains, “AI is really changing how enterprises decide on what content to create. For an AI system, it’s not hard to analyze 10,000 webpages on a topic, compare it with social and search data, and digest it in a second. We can then identify all the connections and then zoom into deeper and deeper levels.”

This helps create a content strategy that is better aligned with how Google ranks websites and helps the business understand the universe of topics they need to cover in order to climb to the top of their market.

Let’s say your company wants to own a market topic like ‘home and garden tips.’ Solutions such as MarketMuse can help identify all the angles you’d want to cover on that topic, showing you the core topics to tackle first and mapping keywords to different user intent profiles.

This is similar to how traditional SEO keyword research works but expands keywords into semantic topics and clusters. So instead of chasing a particular high traffic keyword, you’re mapping content to buyer profiles and models of what Google looks for in that topic.

“If you’re pulling data from Google AdWords and have 300 writers all covering that big list of keywords, you’ll get a lot of duplicate efforts,” Coyle says. “And you’ll miss opportunities to map content back to comprehensive buyer profiles as Google’s AdWords data has a preference for end-of-the-funnel direct response keywords.”

I asked Coyle and Balogh to break down this process into steps. Here’s what they recommended:

  1. Research the entire set of concepts and topics that your business should own in your market. This goes beyond the high traffic keywords that Google’s AdWords data provides. What’s the semantic universe of topics? What are the underlying keywords and user intent profiles that encompass everything your brand should be focused on?
  2. Do an audit of your existing inventory. What are you covering well? What are the gaps in coverage? What efforts are you duplicating between teams, business divisions, or regions? Where are the blind spots?
  3. Look at competitors. What do they cover well? Where will you have a lot of challenge in beating competitors? Where is some low hanging fruit?
  4. What does it truly mean to write great content on this topic? Machine learning and AI can help you analyze the model of authority in this industry, showing you the coverage and related topics you need to cover. What’s the investment you’ll need to compete? What is the model site that Google is looking for to represent this topic?
  5. In the end, you’re looking to answer two questions. What should you be focusing on as a universe of topics and editorial themes? How does that universe relate to your existing inventory of content? For large organizations with thousands of webpages, AI can help to understand the gaps and spot the editorial areas you need to develop.

5. Personalization for everyone

Consumer brands such as Facebook, Spotify, Amazon, and Netflix have given shining examples of how powerful and profitable personalized web experiences can be. AI is helping to make personalization a reality for brands of all sizes.

“You spend less money on marketing, annoy people less, and provide messages that are more relevant and welcome to customers. These are the core benefits of personalization,” Jeff Hardison, VP of marketing at Lytics, told me.

Lytics is a customer data platform. This new AI-powered technology was recently added to Gartner’s “Hype Cycle for Digital Marketing and Advertising, 2016.

Here’s why Gartner thinks this new type of customer intelligence will be critical for personalization:

The customer data platform addresses an acute need for modern marketers are ever in search of that elusive complete view of the customer, beyond the acquisition stage… A bridge between the traditional marketing database or post-sales CRM system and multichannel campaign management execution engines, the customer data platform arose from the need for a solution that could be controlled and deployed by marketers to unify customer identity in a privacy-compliant way, manage first-party data and connect execution across multiple point solutions.

While most organizations likely have personalization and customer data platforms written in a strategy powerpoint somewhere, Hardison believes that a lot of work still needs to be done before organizations take full advantage of the potential of personalization.

A lot of personalization used by organizations is still quite simple. It might be ‘you’ve liked this product and other people like you have bought similar products’ or articles are tagged with things such as email marketing and data science and then the user is shown similar articles based on those tags.”

“But the human brain can only make very simple recommendations,” Hardison says. “For example, you can look at five types of blog posts and say if people are interested in this topic, they’ll likely want to read on this other topic too.”

Machine learning can include much more data in their recommendations, looking at a user’s individual behavior across many different types of content such as reading case studies, clicking on an Instagram ad, opening emails, watching videos, and completing tasks in the product. This creates a much more nuanced type of personalization than human analysis could provide.

One of Lytics’ customers is a major publisher with multiple media properties. They used Lytics to create reader profiles—what has this person looked at on our website? What have they clicked on in our marketing? What is their purchasing history? Do they have a help desk support ticket open? What types of content formats do they typically engage with?

Once they gathered this information, they exported all of these reader profiles to their website tools and then only serve up the type and format of content these profiles engage with.

“This helped them cut down on costs of marketing and deliver more relevant content to their key customer personas,” Hardison says.

“They can then go to advertisers with these say five customer profiles—and say, would you like to reach this type of customer, here’s what they click on, what topics they follow, and the best ways to engage them.”

Again, I wanted to know how organizations might start putting some of this theory into action.

“Look at your existing tools,” he told me. “Your email service provider or website content management systems or ad retargeting tools. A lot of these vendors will integrate with a customer data platform like Lytics. You can then look for ways to add sophisticated personalization to your current campaigns or as a pilot on your website.”

Learn more about AI

Here are a few more resources that explain how AI works and how enterprises can put this technology to work.

How does machine learning work? This visual tutorial shows how machine learning can distinguish between homes in New York from homes in San Francisco.

Gartner recently held a webinar called Cognitive Computing and AI: Smart Machine Technology for Business Success.  Tom Austin, VP and Gartner Fellow, talks about how to set realistic expectations for AI and shares which use cases drive the best business results.

The Marketing Artificial Intelligence Institute is tracking the rise of AI in marketing technology. Their blog spotlights new AI companies and covers how different industries are personalizing marketing campaigns with machine learning.

You can find out more about customer data platforms in Gartner’s Hype Cycle for Digital Marketing and Advertising, 2016.