How One Marketing Team Made AI Part of Its Daily Work

A marketer’s job is never done. Marketing leaders increasingly recognize that their teams need more support to square consumers’ around-the-clock expectations with the bandwidth of human marketers. According to a survey conducted by Forrester, nearly nine out of ten marketers believe their organization must increase its use of AI to stay competitive, especially as resources remain flat or decrease. Yet only half of marketers feel that they have adequately adopted AI, a discrepancy that isn’t necessarily surprising.

AI isn’t just a switch marketers can flip. And there’s often a bit of misplaced shame attached to using it on tasks you could perform yourself. Some ask, “If AI can do this, what value am I adding?” But AI alone isn’t a strategy; it’s a means to an end.

For marketers, turning to AI for help should be as natural a habit as Googling. We can build that impulse by deliberately experimenting with new use cases every day. Marketers of all levels should pause at every turn to ask whether AI could make something faster, easier, or better — and then, experiment away.

I’ve seen firsthand how building familiarity with AI can offer compounding benefits for a marketing team. Wink, the in-house agency at Intuit Mailchimp, deployed AI in 2023 to deliver our most impactful campaign yet — with a spot that tested into the top 5% of Ipsos ads across the world. We used a combination of generative AI and human expertise to collaborate more constructively, to quickly optimize, and to thoughtfully localize our message for various markets. And with the right commitment to experimentation, I believe this success can be replicated. Here’s how.

Show versus tell

Leading up to our fall 2023 campaign, our research confirmed that even advanced marketers were struggling to personalize messaging for customers in a meaningful way, even when they had the data to do so. We knew we needed to communicate the customers’ problem (and our product as the solution) with wit and nuance in 30 seconds or less.

One idea from the marketing team quickly emerged as the best way to convey our audience’s key challenge — that of a physical cluster of customers tangled together in a big, messy ball. This tangle of customers represented a problem we dubbed “clustomers,” or amorphous blobs of undifferentiated customers that all receive the same marketing message. “Clustomers” is what marketers wind up with when they fail to segment targets and personalize their messaging appropriately.

As we started talking about what that “clustomer” could look like, we knew pretty quickly that we had the building blocks of a great idea. But we needed to make sure we were all seeing it the same way. We needed to pull our vision apart and put it back together in order to confidently proceed. So, we turned to prototyping with an AI-powered image generator to help us rally around a shared vision.

The image needed a lot of revisions to display the idea we saw in our heads. Yet at each step, talking about what the AI-generated image got wrong was equally as valuable as talking about what it got right. Whether we were adding props to demonstrate the customers’ shopping habits or adjusting the characters themselves to better reflect our diverse audience, each detail was debated until we could coalesce around a visual that more accurately reflected our imaginary “clustomer.”

AI empowered us to move quickly during this ideation phase because it allowed the team to show versus tell. We narrowed down concepts in one meeting, not the five-plus discussions spread across several weeks needed in the past for various rounds of creative. Prompting together forced us to have the hard conversations earlier. And by discussing intricate depictions rather than loose sketches, the team felt more comfortable committing to the idea wholeheartedly.

Using AI to help express ideas can increase not only your team’s velocity, but its shared understanding. When we can quickly bring concepts from our imaginations into a shared plane, we empower groups of all sizes to collaborate around the same vision. Doing this daily, on projects big and small, can drive momentum and confidence, and free up resources for more advanced strategic adjustments as you bring your idea to market.

Optimize with speed and precision

During research and testing, we continued to capitalize on our experience prototyping with AI. First, we used generative AI to create variations of the “clustomer” scripts that riffed on different digital marketing concepts — such as abandoned carts or varying discounts — to see which landed best. Then, as we spoke with focus groups, we further refined our vision.

We started, for example, with a “clustomer” made up of many different women, all wearing blue dresses — a visual representation of exactly the kind of seemingly specific yet ultimately meaningless customer profile many e-commerce marketers rely on for subpar segmenting. But we quickly found that we didn’t need to be quite so niche in order to drive home the message about personalization at scale; a cluster of customers who had hardly anything in common was actually much more digestible.

Previously, bringing concepts to a focus group didn’t allow for much adjustment between sessions. At best, you could change a script on a slide. Because generative AI could take our feedback and drastically change imagery, scripts, and more in just minutes, we could use the insights that we heard in one session to adjust our prompts and generate new material for the next.

This task required human intuition and machine speed. We relied on our team to interpret micro-expressions, extract practical meaning from verbal feedback, and map out adjusted prompts based on what we heard from a focus group. Then, we relied on generative AI to implement that feedback immediately for an improved prototype. Fluency with both AI and with people — two separate sensibilities, both strengthened through experience — allowed us to retest our concepts with optimizations within minutes. Ultimately, we shared an extremely specific vision with our producers for the final spot (a project for which we enlisted a human team of creatives).

We came to think of AI like a new colleague: Getting acquainted through smaller projects and seemingly trivial interactions on a daily basis can teach you how to better communicate. It’s still your responsibility to clearly distill what you need, and to deliver coaching and feedback when you don’t get it. But the more you work together, the more capable you become at requesting the changes needed to land on your desired business result quickly.

Personalize with soul

Mailchimp has customers in more than 190 countries, and many of them were experiencing the same “clustomer” problem — even if they wouldn’t all describe it as such. We needed to localize our message, but we couldn’t just translate “clustomer” — our combination of two similar-sounding English words. Despite all the ways AI aided us throughout this campaign, human ingenuity once again led us forward.

We enlisted creatives on the ground in our target markets to workshop copy adjustments, testing multiple unique combinations of Spanish words, for example, to ensure our “clustomer” had the moniker that best endeared it to audiences in Spain. We also consulted bilingual marketing experts both in-house and outside of our organization to confirm that we had continuity between our original concept and the localized product.

Still, the AI in our operating mechanisms allowed us to get the optimized campaigns to market faster, whether we were adjusting spelling for different English-speaking markets or relaunching in a new language. In some cases, we relied on AI tools to provide rough translations as a starting point for alternative scripts. In others, we leveraged creative automation tools for versioning, adjusting for different asset sizes and types, to more quickly produce campaign materials across channels. Turning to AI for something as simple as optimizing banner ads across platforms can help maintain critical momentum as you take a campaign to market.

With every use case, there are limits to automation. Experimentation is one way to confirm these limits. Deploying AI tools in service of efficiency may not be the right move for your team in every situation, whether you’re creating something new or personalizing for different audiences. But, on the other hand, if versioning required each individual campaign asset to be re-created manually, many teams wouldn’t have the resources to develop a personalized message for those unique audiences at all. Some balance is needed.

To truly connect with customers, you need a combination of human expertise and AI to ensure sincere personalization. It’s the soulful subjectivity of a marketer coupled with the heightened capacity of AI that will drive deeper connections with your customers.

Start experimenting and stay curious

For a project with such exceptional results, the “clustomer” campaign didn’t have an unusually long timeline or outsized resourcing. We simply deployed the latest technology efficiently, refining and adjusting as we learned in-market. We were able to do this confidently because we didn’t wait for a major project like the Clustomer campaign to build AI into our routine. We just got started.

You can start building AI into your routine, too. Use an image generator to liven up brainstorm meetings. Try out a predictive analytics algorithm to anticipate your customers’ questions and craft relevant content based on your daily web queries. Experiment with an AI notetaker on your next conference call. Or ask a chatbot to summarize the LinkedIn post you’ve drafted, to see if your key message is as clear as you think it is.

Familiarizing yourself with AI doesn’t always have to happen on the clock, either. My son loves science, for example, and we’ve started prompting together to get “ten-minute stories for a 9-year-old” about his latest favorite topic before bed. The more you practice using AI, the more you’ll find ways to harness it.

. . .

When paired with thoughtful human guidance, AI can be an equalizer and an accelerator. It can enable creative processes to be more specific, more provocative, and nimbler. It helps us to quickly and repeatedly interpret data, guiding our actions for maximum impact with minimal resources.

When our teams experiment with AI and share use cases with colleagues, we generate a shared courage and fluency that compounds over time into a meaningful competitive edge. A single experiment each day can build on itself into hundreds of hours saved, game-changing ideas embraced, and a more constructive communication style among teammates. When done right, leveraging predictive and generative AI is thoughtful, collaborative work — and it can deliver the results marketers need to achieve success and remain competitive.

You just have to start.

Copyright 2024 Harvard Business School Publishing Corporation. Distributed by The New York Times Syndicate.

Explore AAPL Membership benefits.