3 Key Ways Marketers Can Use Artificial Intelligence IRL

At Red Rocks in 1983, Bono famously said: “There’s been a lot of talk about artificial intelligence … maybe, maybe too much talk.” (All right, I may have paraphrased that a bit.)

But there has been way too much talk. Talk about why marketers should be using it. Talk of the benefits of AI and machine-learning (ML) technologies in campaigns. Talk about how they’re going to change the face of marketing forever.

According to a recent CMO from Adobe piece: “AI and ML have the irrefutable ability to transform the massive amounts of data being collected from an increasingly connected world into effective insights.”

Yes … but HOW?* Enough with the talk. The new year fast approaches. It’s time for action.

If you’re in the beginning stages of building your 2021 marketing plan, and curious how AI and ML could help support it, here are three practical applications that IMA has started to test — all of which show great promise:


Since it’s impossible to read, qualify and quantify every comment posted on your brands’ digital platforms, the automatic gauging of “sentiment analysis” has become an intriguing marketing insight. Sentiment analysis is a way of identifying and classifying subjective opinions from text; in short, it’s a short cut for learning how your audiences feel about your brand and products, en masse, from their comments.

For example, if a customer writes that “ABC Company has the worst products I’ve ever used,” that would obviously be classified as a negative comment. Whereas, a comment like, “ABC Company just made the Fortune 1000 for the first time, on the strength of record sales last year” would be classified as positive (due to words like “strength” and “record sales”). With enough data, and enough time, machine learning can help a brand crack the code of your audiences’ word choices and get a clear picture of how they feel — about products, company decisions, people behind the brand, etc.

Meltwater is a reporting and analytics tool that provides deep insights into a brand’s social media activity, press coverage, online mentions, and more. And it uses AI to power a number of these functions. With regard to sentiment analysis, Meltwater’s platform has been powered by ML for more than 10 years. And with AI, time is currency.


As we all learned in Marketing 101, a great campaign comes down to four things: (1) audience, (2) channel mix, (3) offer, and (4) creative/message. That last bit — #4 — has largely been treated as more art than science. Smart marketers look at historical data to see what creative has performed best with their audiences. They’ll A/B split test different messages in advance of a campaign, run them for a short test period, and see if one outperforms another before they put their full marketing spend behind the “winner.” This is all sound practice.

But AI can take it to another level.

Platforms like Persado and Pattern89 are driven by ML technology, which takes your creative content — a social media post, display ads, web page, etc. — and deconstructs it down to its components (e.g., headline, image, body copy, CTA, etc.). Then, it taps semantic databases to review millions of similar messages, unearthing the best-performing ones for your specific audience.

Using those data-driven insights, the platform recommends different permutations of your original message — ones that have the best chance of outperforming your original draft. And you can run loads of multivariate tests to find the best performer. That way, you know which message will get the most bang for the buck BEFORE you spend the bulk of your media budget to support it. Pretty nifty.


Some emerging AI platforms promise to automate almost the entire digital marketing process. A marketer can start building audiences by punching in firmographic data (e.g., industry, revenue, employees, NAISC code, etc.) … then connect your CRM, to provide platform data on your existing prospect and customer lists … then plug in your website analytics and social media channels, so the platform can see what kinds of content your audiences are engaging with, and why.

From there, you can start launching sample campaigns — across social, display, paid search and more — and let the platform’s AI suggest and test various permutations of the original. (Similar to #2, above.) The big difference here is that, once the ML starts establishing performance, these platforms automatically downplay the weaker permutations and start pushing more media in support of the winners.

Two such platforms that have caught our eye are Metadata and Albert. And while both seem really promising, they come with some challenges and limitations:

> These platforms don’t integrate with all CRMs & marketing automation platforms. If you’re using SalesForce and Pardot, great. (Or Apptivo and Eloqua, etc.) But if you’re using something a little off the beaten path, you may be out of luck.

> The AI is designed to optimize toward revenue outcomes. This sounds great on the surface. But consider: Such a platform will likely choose Message C over Message F if the former drives opt-ins who deliver higher sales. But this puts you in the last-click-attribution trap. What if Message F’s leads drive higher sales during the next campaign? Or yield a higher lifetime value? Or, what if you’re trying to measure other KPIs along the sales cycle: like engagement, return views, brand ambassadorship, etc.? In short, these platforms can encourage short-sightedness, which is especially dangerous if you take your hands off the wheel and let them do all the adjustment automatically.

> They tend to be expensive. To give you an idea: Using Metadata’s platform, your costs are $4,000/month (after a $2,500 “onboarding” fee) if you’re total monthly ad spend is $10K. An automated platform that costs 40% of your ad spend feels a little rich. (Though you pay the same price for ad spend up $25K/month, which would bring it down to 16%.) Now, for a small in-house marketing team that wants an off-the-shelf, fully automated solution, this could make sense. But for larger teams with more complex marketing operations — that might look at such a platform as supplemental or experimental — the cost feels a little high. Food for thought.

– – – – –

We hope this gives you some practical and useful advice on how you can start incorporating AI and ML in your 2021 marketing plan.

If you’d like to speak with an IMA principal on what tools are best suited to support your specific marketing goals, please reach out to us.


#     #     #


* BTW, any marketer who is leveraging AdWords or the Display Network — and this should be EVERYONE — is indirectly using AI through Google’s TensorFlow project. TensorFlow also leverages AI to inform and drive a host of other well-known products and platforms, like Google Image Search, Google Translate, Gmail, etc.

How can we help you?