By Marcia Riefer Johnston published November 20, 2017

Scale Your B2B Content with Artificial Intelligence: Ideas and Tools Marketers Can Try

scale-b2b-artificial-intelligence-tools-ideas-marketers

“Same house, right?” The question came via Facebook Messenger from a friend who was coming over for dinner. Under his message appeared two options: yes and no. With one touch of a fingertip, my answer appeared in a blue bubble as if I had typed y-e-s myself.

That experience, a first for me, was so logical, natural, convenient, and simple that I hardly noticed it. An app had recognized my friend’s message as a yes-no question and had presented me with ready-to-use replies. Nothing about the exchange shouted, “Hey! Check it out! Artificial intelligence at work!”

Only after I heard Paul Roetzer’s Content Marketing World talk did I realize that my experience represented exactly that: artificial intelligence at work.

In fact, artificial intelligence is at work all around us. And this “science of making machines smart” (Paul’s favorite definition, which comes from Demis Hassabis, co-founder and CEO of Google DeepMind) is beginning to open possibilities for B2B marketers who want to increase efficiency, boost performance, and create a competitive advantage.

If you take away only one thing from this post, let it be Paul’s mantra: “Try it!”

If you take away two things, let the second be “Don’t wait!”

Try #artificialintelligence in your B2B marketing and don’t wait, says @PaulRoetzer. #intelcontent Click To Tweet

This post covers highlights of Paul’s CMWorld talk, Machine-Assisted Narrative: How to Transform and Scale Your B2B Content With Artificial Intelligence.

Artificial intelligence: Part of everyday life

As a marketer, you’ll never have to understand artificial intelligence in depth. “You don’t have to know how it works. That’s irrelevant to you. You just have to know that there are AI-powered tools that do things you weren’t capable of doing before that can now start to inform your strategy,” Paul says.

For the record, he notes that artificial intelligence is an umbrella term for lots of related terms, including machine learning, deep learning, natural-language processing, natural-language generation, computer vision, and image recognition.

ai-machine-learning

Image source

AI is part of everyday life in ways that you may not recognize. Examples:

  • Photo applications, such as Facebook and Apple Photos, outline people’s faces and prompt you to tag them by name.
  • Google and other search engines suggest text strings that you might have in mind as you type, and personalize your search returns based on your location and your past online behavior.
  • Phone and texting apps offer context-appropriate quick replies (like “I’m in a meeting”) for you to tap.
  • Netflix recommends shows based on your viewing history.
  • Amazon recommends products based on your buying history.
  • Online media, such as The Washington Post, recommend articles based on your reading history.

You get it. As Paul puts it, “Your life is already machine-assisted.” Less apparent is that your marketing will be too, Paul says.

How exactly? No one knows yet, but Paul has some ideas. Imagine, for example, you, as a content marketer, set up a system that knows enough about each visitor to surface precisely those nuggets of content from your site that make that person’s day. What would happen to your conversions?

Caveat marketor

Caveat marketor. That could be Latin for best get ready. While no one knows the impact AI will have within marketing, inevitably some jobs will disappear and others will emerge. Knowledge work – your work – will be disrupted, Paul says.

No one knows the impact #AI will have. Some jobs will disappear & others will emerge. @PaulRoetzer Click To Tweet

Some big brands have high expectations of what will come of applying AI to activities that humans are paid to do today. For instance, according to one report Paul points to, Coca-Cola is “ditching flesh and blood creatives in favor of software algorithms in an experiment to see whether AI bots have what it takes to beat their human masters.”

Similarly, in October 2016, the lingerie retailer Cosabella replaced its digital agency with an AI platform named Albert. “It has more than tripled its ROI and increased its customer base by 30%,” as reported in this article in March 2017. The brand claims that it “will never go back to humans.”

.@shopcosabella replaced its digital agency w/ an AI platform & more than tripled its ROI. @CampaignLiveUS Click To Tweet

So yes, caveat. Beware. But don’t panic. Paul suggests looking at the situation this way:

For the foreseeable future, humans are uniquely capable of creative, emotion-based functions. If it’s a task that is data-driven or can be executed by defining a set of rules, a machine will eventually outperform humans at it. All it takes is for someone to have the vision, desire, ability, and funding to build it.

In short, take the initiative to look for ways machines can help marketers do what machines do best, and keep doing the things that humans do best.

How to get started

As you review all the possibilities mentioned below, you may be overwhelmed. Where’s a marketer to begin?

Paul’s advice: Pick one use case, a task that eats up a lot of time, some administrative or tactical thing your team hates to do. Assume a tool either exists or is being built that can enhance what you do in that area. Get a basic understanding of what’s possible. What you learn “will give you superpowers that others don’t have,” Paul says.

If you pick a couple of tools and start doing some of these things, no one’s going to understand what you’re doing. It’ll be magic to people in your company. It can be your competitive advantage.

As you read this article, look for an idea that has your name on it – the one thing you’d like to try.

Experiment with #AI to automate your most time-consuming #marketing tasks, says @PaulRoetzer. #intelcontent Click To Tweet

An AI framework for marketers: The five Ps

Paul, who created the Marketing Artificial Intelligence Institute about a year ago, has developed a framework of five Ps:

  • Planning
  • Production
  • Personalization
  • Promotion
  • Performance

This framework outlines ways that marketers might take advantage of AI today and in the future.

Planning

Planning, the first category of the AI framework, relates to marketing activities like predicting consumer behaviors, defining strategies, prioritizing activities, and determining allocation of resources.

In this area, not much AI technology is available yet.

Topic clusters

One example of AI-supported planning is HubSpot’s ability to build topic clusters (as shown below), which gives content teams a way to discover topics they might want to write more about.

Topic clusters give #content teams a way to discover topics they might want to write more about. @PaulRoetzer Click To Tweet

ai-planning-hubspot-topic-clusters

Image source

Monitoring competitors’ digital footprints

You can support your planning by using tools like Crayon and Pathmatics (shown below), which enable you to monitor the digital footprint of your competitors, giving you information to use when forming your content marketing strategy.

Tools like @Crayon ‏& @Pathmatics enable you to monitor digital footprint of competitors.‏ @PaulRoetzer Click To Tweet

crayon-pathmatics-planning-tools

Analyzing content

Marketing planning often involves analyzing content as a basis for making decisions about future content. One free AI tool anyone can use for this purpose is IBM Watson Analytics. You import any dataset – for example, a CSV file – and explore it.

Paul gave IBM Watson Analytics a try. He used BuzzSumo to export analytics data from the previous 12 months on CMI’s website: titles, URLs, total shares, word count, etc. He hit export, cleaned up the spreadsheet a bit (shown below), and imported it into IBM Watson Analytics.

buzzsumo-ibm-watson-analytics

Click to enlarge

“I don’t know how it works, but I know it’s an AI tool,” he says. “I just threw the spreadsheet in there and watched what happened.”

Paul typed in “total_shares,” and Watson Analytics came up with a set of questions that it predicted people might want to ask (shown below).

ibm-watson-analytics

Click to enlarge

“Instead of me trying to figure out what to do with this data, Watson recommends things to look at, questions to ask of this dataset,” Paul says. “I don’t have to be a data scientist. I don’t have to be an analytics expert.”

From the set of generated questions, Paul picked, “What is the breakdown of total_shares by author_name?” – a question he hadn’t even considered asking. Instantly, Watson Analytics delivered the breakdown in a stunningly simple visual: a bunch of blue boxes, one per author, packed into a big rectangle in order of the number of times that author’s CMI posts have been shared in the past year (shown below).

watson-analytics-breakdown-total-shares

Click to enlarge

At a glance, Paul could see that nine authors – those whose boxes take up the left half of the big blue rectangle – account for about half of all shares of CMI content. (Whoa … there’s my name, fifth. Thanks, all of you who have shared my posts. And thanks, Watson Analytics, for the rush.)

Paul suggests that content marketers get out of their comfort zones and try tools like BuzzSumo and Watson Analytics to increase their insight into questions like: How are your authors performing? How do your articles perform based on word count? How does your content perform within specific channels?

Content marketers: Use tools like @BuzzSumo & @watsonanalytics to increase insight. @PaulRoetzer Click To Tweet

“I imported a free dataset into a free analytics tool and used AI to help me crunch this,” Paul says. “It’s a simple use case with tools that are readily available to anybody.”

Production

Production, the second category of the AI framework, relates to marketing activities like these: creating, curating, and optimizing content, including blog posts, emails, landing pages, videos, and advertisements.

In this area, “we’re not that far along,” Paul says. Yet he touches on several production-related use cases, as described here.

Curating content

Tools use AI to help people surface the right content to share across a network or to enhance a blog. You may even be using a curation tool – Curata (shown below) or Scoop.it, for example – and not realize that you’re using AI.

Use content curation tools like @curata or @scoopit to surface the right content to share, says @PaulRoetzer. Click To Tweet

curata-ai-tool

Image source: Curata

Writing email subject lines

One new tool – Phrasee – creates email subject lines that can result in more opens, clicks, and conversions than subject lines written by people. This software uses AI to evaluate email content and recommend 10 subject lines based on scores (as shown below) that indicate how each is likely to perform.

Use @phrasee to create email subject lines that can result in more opens & conversions. @PaulRoetzer Click To Tweet

phrasee-tool

Image source

A company video says that Phrasee “understands the emotions, sentiments, and phrases that resonate with your audience. Your past, present, and future results drive advanced algorithms that generate better subject lines than humans can, all tailored to your brand’s voice.”

Paul urges you to compare how the machine-generated subject lines do against your own in A/B tests. “Try it,” he says. To listen to Paul is to hear those two little words over and over.

Analyzing text

Tools that analyze text for grammar, sentiment, style, and tone of voice include Acrolinx, Grammarly, and AtomicReach. Acrolinx, which has been doing this kind of AI for enterprise companies since 2004, offers software that integrates into authoring tools and can assess content in multiple languages against organization-specific guidelines.

Use tools like @Acrolinx, @Grammarly, @Atomic_Reach ‏for grammar, sentiment, style & tone. @PaulRoetzer Click To Tweet

acrolinx-tool

Image source

Creating data-driven content

Data-intensive content (based on financial info, analytics, product info, etc.) lends itself to automation. A human creates a template, and machines do all the rest at scale, spitting out updates as often as you please. While fill-in-the-blanks automation isn’t AI – the computer isn’t learning as it goes – Paul includes it in his AI talk as a step in that direction.

Here’s an example of this type of natural-language generation (NLG) as executed using the Automated Insights tool called Wordsmith, which companies in over 50 industries used to generate over 1.5 billion “NLG-powered narratives” ­– reports, articles, etc. – last year.

automated-insights-wordsmith

Image source

This approach strikes me as a Mad Libs®–type exercise without the “mad.” A number here, a noun there … voilà! Natural-sounding sentences and paragraphs.

Automate your reports Mad Libs® style using @AInsights #Wordsmith, says @MarciaRJohnston via @PaulRoetzer. Click To Tweet

It was this type of automation that inspired Paul in 2012 to dig into the possibilities of AI for marketers. He was attending a conference where the managing editor of The Associated Press and the CEO of Automated Insights were on a panel. They told the story of how The Associated Press had automated its earnings reports using Automated Insights tools. They went from humans writing 300 earnings reports per quarter to machines generating 3,000 reports per quarter.

“I sat there as a content marketer at heart and thought, ‘This could change everything,’” Paul says. “Can you actually write content with machines?”

He took on that question himself. His agency now uses Automated Insights with Google Analytics to automatically generate most of the text for their clients’ website reports. Where his team used to spend eight hours creating each report, it now simply adds insights and recommendations to the generated report, which takes less than an hour. The automated approach has cut analysis and production time by more than 80%.

Recognizing and auto-tagging images

In a tool like Clarifai, an algorithm assesses images ­– for example, items of apparel ­– and tells how likely each image is to be one type of thing or another. The example below shows that Clarifai is 97.9% sure that the image is of a turtleneck. As humans work in the background to confirm or correct the machine’s guesses, the machine gets more accurate.

clarifai-tool

Image source

Personalization

Personalization, the third category of the AI framework, relates to the one-to-one tailoring of people’s experiences through intelligently automated content, such as emails, product recommendations, web content, and augmented-reality and virtual-reality experiences.

Paul calls this the “Amazonification” of content marketing. This is probably where most of the money has gone so far, he says.

Like data-driven content, personalized content doesn’t always qualify as AI. Still, Paul talks about personalization in the context of AI for marketers because automation is a big part of content’s evolution. And personalization sometimes contains elements of AI, with the machine getting smarter on its own without a human telling it how to do it.”

The NBA basketball team Orlando Magic, for example, personalizes some of its emails. The example below shows an email to a season ticket holder who was reselling tickets. It was created and sent by the Wordsmith tool; no human was involved (beyond setting up the rules). Basically, the message says, “Your tickets aren’t going to sell tonight. You can either waste them or trade them in for Magic Money.”

orlando-magic-wordsmith

Image source

The personalized emails are “boosting fan engagement and season ticket renewals,” Paul says. Those results spell real money.

Promotion

Promotion, the fourth category of the AI framework, relates to AI for social media. Paul describes it as managing cross-channel and cross-device promotions to drive engagement and actions, including audience targeting, social publishing, and management of digital paid media.

In this area, too, Paul says, not much is happening yet, although a lot of companies are experimenting.

Cortex is one such company. Its software, also called Cortex, makes recommendations for content headed for social media. (Like many of these startups, Cortex comes out of Boston. “MIT has a little bit to do with AI development right now,” Paul says.) This software, according to a company video, helps marketers create the images, text, and videos that “inspire consumers to take action” yielding a 40% to 500% increase in marketing results while saving an average of 8.5 hours per week.

.@meetcortex is a tool that makes recommendations for #content headed for #socialmedia, says @PaulRoetzer. Click To Tweet

Cortex suggests colors, hashtags, keywords, image types, and publication dates and times. You can set it up to make these choices automatically.

cortex-tool

Image source

Performance

Performance, the fifth category of the AI framework, relates to marketing activities – activities like turning data into intelligence through automated narratives and insights and then using that intelligence to optimize performance.

Today’s tools are “not very smart yet” when it comes to reporting on performance, but they can answer questions like how was traffic in August.

Paul predicts the tools will get smarter faster, especially since Google has jumped in with its recent release of Google Analytics Solutions. If you have a list of questions you want answered every month, you can set up this tool to ask Google those questions (as shown below) instead of asking a data analyst. “Google’s probably going to tell you better than the data analyst – and instantly,” Paul says.

google-analytics-solutions

Image source

While we’re on the subject of AI and performance, think back to Paul’s story, told earlier in this article, of his agency automating its clients’ analytics reports ­– cutting its analysis and production time by more than 80%, from eight hours to less than an hour per report. Where could you use that kind of boost in performance?

Conclusion

You don’t need an AI algorithm to conclude that Paul’s main message is “Try something now.” The technology may be young, but it’s evolving fast. Don’t wait.

Which AI tools have you and your teams tried? What have you learned?

Here’s an excerpt from Paul’s talk: 

Please note: All tools included in our blog posts are suggested by authors, not the CMI editorial team. No one post can provide all relevant tools in the space. Feel free to include additional tools in the comments (from your company or ones that you have used).

Want to view Paul’s Content Marketing World presentation in full to learn about more tools and examples, and watch more presentations from hundreds of other experts from the 2017 event? Secure your video-on-demand subscription today.

Cover image by Joseph Kalinowski/Content Marketing Institute

Author: Marcia Riefer Johnston

Marcia Riefer Johnston is the author of Word Up! How to Write Powerful Sentences and Paragraphs (And Everything You Build from Them) and You Can Say That Again: 750 Redundant Phrases to Think Twice About. As a member of the CMI team, she serves as Managing Editor of Content Strategy. She has run a technical-writing business for … a long time. She taught technical writing in the Engineering School at Cornell University and studied literature and creative writing in the Syracuse University Masters program under Raymond Carver and Tobias Wolff. She lives in Portland, Oregon. Follow her on Twitter @MarciaRJohnston. For more, see Writing.Rocks.

Other posts by Marcia Riefer Johnston

Join Over 200,000 of your Peers!

Get daily articles and news delivered to your email inbox and get CMI’s exclusive e-book Get Inspired: 75 (More) Content Marketing Examples FREE!