AI reshapes agency economics, but their contracts are still scrambling to catch up

Increasingly, agencies are using AI tools to automate more parts of the marketing workflow. They’re also peddling proprietary generative AI tools to woo clients, win new business and compete in the AI arms race. The ripple effects are seen in how accounts are staffed, how work gets produced and the cost structures of that work — all of which are typically listed out in Master Service Agreements (MSAs). 

“Ultimately, people want disclosure. They want to know what tools are being used,” said Keri Bruce, partner and head of the advertising group at the international law firm Reed Smith. 

Piecemeal approach to AI clauses

Rather than automatically updating each client MSA, agencies are instead revising agreements by way of addendums, clauses and added on documents. There are a few instances that trigger those changes: a client specifically requests a seat within an agency’s proprietary AI tool, a client asks that specific AI language be added to the agreement or a new AI-powered tool is being used for a client-specific project. 

“We are not yet to the point where we’re creating appendix A or B specifically to spell out agentified execution,” Scott Shamberg, president and CEO of Mile Marker agency, told Digiday. “We are at the point now where we are just simply working that into existing MSAs.” Meaning, the agency isn’t creating separate extra documents for AI projects, because the rules and guardrails are ever-changing on a client-by-client basis, Shamberg said. Instead, those matters are being built into existing agreements.

Addendums, clauses and contracts

At Mile Marker, contracts are tweaked to address how metadata is managed – who owns the IP of AI-generated content and transparency around what tools are being used. Those tweaks show up by way of clauses in the contractual agreements, Shamberg said.

Over the past six months, Go Fish digital marketing agency has added specific clauses to MSAs around things like brand safety and data protection, according to David Dweck, Go Fish president. Go Fish does not typically charge for its proprietary tech.

Crispin has updated contracts incrementally, adding addendums to MSAs when needed, Freddy Dabaghi, Crispin’s chief transformation officer, told Digiday. 

Clients are keen on understanding how their data is being stored and used in these AI environments, per the execs — especially in the walled garden environments that proprietary generative AI tools can create. Clients are also looking to have agencies spell out details about human oversight, indemnity, disclosure of tools and intellectual property in the agreements.

Evolving business models

For agencies, the priority is to keep work flowing with a tweak-as-necessary approach to MSAs rather than an overhaul. On the one hand, it’s to account for clients’ varying comfort levels around AI, risk thresholds and internal rules regarding data usage. And because AI is moving faster than agency contracts can keep up — both in terms of tech and cost structures. 

“We’d rather stay with what’s familiar and how advertisers are paying agencies for a century versus trying to change the game up by trying to be a SaaS company,” said Dweck.

Historically, agencies have operated on a traditional business model charging clients hourly rates. Now, agencies are increasingly relying on AI tools, thus reducing the amount of hours required for work and ultimately upending the agency business model. 

To keep dollars flowing, agencies are rolling out proprietary generative AI tools and, in a way, pivoting toward a software as a service business model. That changes the payment structure, which agencies are still trying to determine.

“A lot of times it becomes a function of time and money,” said Brian Yamada, global chief innovation officer at VML. Until the industry can agree on an AI cost structure, it’ll be difficult to build out standard contractual agreements that include those changes, Yamada added. 

“The market is changing so quickly that at least my advice is to make sure you’re building some flexibility into that, to re-examine,” he said.