Inside agencies' AI practices and how they're working with clients

Within the last six months, agencies have gone from experimenting with AI tools to launching AI-specific offerings and services focused on consulting, strategy, creativity and other areas. These services are being offered by new entrants into the space as well as agencies that were already frequent users of AI technology such as ChatGPT, Midjourney and DALL-E 2—before those tools piqued mainstream interest in AI. Now, these shops are thinking about the way they frame their AI services to existing and potential clients. AI marketing blog How brands are using ChatGPT, DALL-E 2 and other tools Read more here Here are five ways agencies are doing so. As a way to do more with less Dept, a user of AI for more than a decade, launched an AI practice in mid-February, according to Isabel Perry, VP of emerging technologies at the agency. The practice includes four leaders chosen from across the agency's disciplines globally, 40 data scientists and engineers, and utilizes the expertise of 400 Dept employees across various data, creative, growth and engineering teams. Some previous uses of AI from the agency include a machine learning hand-tracking system to teach the American Sign Language alphabet and an AR experience for eBay’s MCM Comic Con campaign that featured generative AI collectible cards. “In the last 12 months, the power of AI sort of moved out of the domain of just engineers and developers and data scientists and into everyone's hands,” Perry said. In the last year, Dept has seen an increased interest from clients in wanting to learn about these AI tools and how they can help the agency do “more with less,” according to Perry. Also read: AI copyright ruling could pose issues for agencies The agency felt the need to formalize its offering, said Dimi Albers, global CEO of Dept, because “there's always that moment where you need to tell the story more broadly.” Last year, 30% of the agency’s revenue came from AI-enabled work, which included using the tech to improve the spend management of a certain media client. In one case, Dept worked with a beauty retailer to help target its consumer messaging based on their online shopping habits. The agency handles over $3.5 billion in media budgets for clients, 70% of which implements some type of AI, according to Albers. The goal for the agency is to have 80% of its revenue come from AI-enabled work by 2025. “In the short term, helping clients understand how we can help is incredibly important, but for long-term growth, it's as important to say, 'Here are the people you need to work together with and build the company of the future now,'” said Perry. Also read: 13 marketing leaders to know Consulting on strategy AI-powered agencies such as Supernatural and more recently Product have popped up over the past couple of years. In November 2022, Supernatural launched an AI consultancy practice after seeing growth in that area following the agency's 2021 launch. Currently, nearly half of the agency’s revenue comes from consulting, according to Chief Strategy Officer Mike Barrett. As the need for consulting work grew, the agency saw the need to launch the discrete practice. “The types of questions that management consultants tend to answer are 'What drives growth in my category? How do people make decisions? And where are the most valuable audiences? Are they the people I have now?'” said Barrett. Also read: A generative AI guide for brands “We are getting engaged to answer those questions,” said Barrett, who previously helped lead Heat, an agency acquired by consulting giant Deloitte in 2016 for an undisclosed cost. “The reason we split it off as a consulting offering is because we want clients to know that they don't have to hire a whole advertising agency to work with us on this stuff. If they want us to come in, bang out some smart stuff, and hand it over to an internal team, we're happy to do that.” Its consulting practice can solve issues beyond consumer marketing, according to Barrett, noting that Supernatural is helping a hospital client work on employee retention. The agency is building an operating system that encompasses different AI tools to generate creative assets and brainstorm strategically. To inform strategy decisions, clients answer 39 questions in the operating system that then uses audience data and other data to generate different strategy frameworks. Currently, the agency is licensing its system to a small number of agencies and considering expanding that on a larger scale. Researching creative and new product development Product, which uses AI for all its projects, noticed increased client interest specifically in using AI to test creative. This led to the launch of its new offering called AI Studio. “We’ll use AI to create very rapid, low fidelity versions of ads that illustrate [several] different messages,” said Founder Aaron Shapiro. “We'll use AI to then create different types of creative approaches, [such as] using people or not [in the ads] or other different types of derivatives. Then those low-fidelity ads, which are basically photos with copy on it, can then be run very inexpensively as test campaigns on social where we test how well those different creative messages are doing with real data.” From there, the agency can turn the ads that work into higher-quality ones. This system can also be used to test out ideas for a TV spot, for example. Shapiro said the agency recently worked with a CPG brand to test which product line out of several the company should launch. The test, which included 100 creative assets, was completed in three weeks from the moment the agency was hired and the product that received the most engagement was launched six months later, according to Shapiro. “It's like hiring a research firm,” Shapiro said. “You give us the set of messages and the creative approaches that you want to test, and then we go ahead and test it and then we charge a fee for those several weeks of activity.” Check out the latest Ad Age events and award programs To manage a brand's reputation In January, Boston-based independent Boathouse launched its first AI tool called “Narrative Transformation.” The tool uses technology from software company NetBaseQuid and media monitoring company Signal AI to gather data from websites such as Reddit, Twitter, Google News and Glassdoor to understand public sentiment through trending hashtags, key terms or employee attitudes. The University of Massachusetts Global is using the tool, which is still in beta to get a better understanding of how people perceive the university and how it compares to other online universities, according to a statement by the agency. Boathouse President Peter Prodromou said that using AI this way is more “actionable” than the use of generative AI. Also see: How AI text-to-video could impact marketing “Our best value model is in using AI to inform the portfolio mix of services [for a brand] as well as the messaging, narrative and the overall structure of a campaign,” Prodromou said. “You'll see us more and more going into every pitch with some level of underlying AI output to inform where we're going.” While the agency is using the tool for some consultative purposes, the ultimate goal is to use it to gain a larger share of a company’s advertising business. “There are instances where we are using it to help a CEO to better understand the performance of their existing marketing infrastructure,” Prodromou said. “That's not the ideal scenario. As an agency, you want to sell the whole thing. And I think you'll see few instances of us just selling it as a one-off.” Since the tool can be used to identify patterns and identify quantify categories such as “passion” for a brand, Prodromou sees the potential of using it for crisis management before an issue becomes a threat to a brand. “Social media gives people a voice and with that voice, you can start to see where a thread is going to start to emerge" or "where there are conversations that are starting to happen that are negative about the performance of a product,” Prodromou said. To position themselves for a new content creation world PJ Pereira, creative chairman and co-founder of Pereira O' Dell, became interested in AI seven years ago, when he began research for a book he was writing that has an AI-based character. Now, Pereira O' Dell just hired its first full-time AI artist as part of its AI Lab, an internal team of four employees dedicated to experimenting with generative AI tools like Midjourney. The task for the lab is simple: Finish one experiment a week. So far the agency has created AI videos from prompts such as “Joker in a world of knitting” and “Pulp Fiction in an Atari world.” These projects are purely experiential to learn about these generative art tools, according to Pereira. “This is the time to play, not to offer, so that's why AI is not an offering that we put within the agency—it’s a lab,” Pereira said. “I wanted to do that without having the constraints of a brief, without the constraints of generating revenue. If eventually this process leads to an idea that I can bring to a client, and that can become a project, awesome.” Though the end result isn't perfect, he said that the Joker video, without the use of AI, could have taken months to make but instead took hours. Recently the agency was able to revive a project idea for a client that was previously considered too expensive but was made more cost-effective with generative AI tools. Pereira believes that the AI artist role will at some point be institutionalized in other agency departments including social, design, and production. To hire the agency's first AI artist, he scoured communities across Instagram, LinkedIn, and Facebook and asked promising candidates to participate in a brainstorming session. He said that working on a project both parties are invested in is a more telling experience than a formal interview for this type of role. “The paradox of this is, the more technology takes over, the more important the human connection becomes,” he said.

Inside agencies' AI practices and how they're working with clients

Featured Stories