Generative AI has reached critical mass among marketers and advertisers, from ideation to content creation. What has not caught up is measurement. For an industry obsessed with ROI, few businesses have built straightforward frameworks for tracking what their AI investments actually return.

CMOs currently measure AI benefits across a scatter of metrics: time saved, payroll cost reductions, consumer sentiment. Each has pros and cons, and few apply across an entire business. No standard dashboard has emerged.

"We do have measures that give us some sense of how effective it is," said Cheryl Guerin, evp of brand strategy and innovation at Mastercard. "I haven't been able to say, 'Oh, I've saved X amount of money because [of AI].'"

Without a common way to back up AI's value, marketers are taking a more cautious approach. They operate within the bounds of what is provable, not what is possible.

Dashboard signals

For Bryan Demaranville, CEO of L.A.-based jewelry brand Melinda Maria, the scoreboard comes down to payroll and customer service. He estimates AI-powered tools generate $1.2 million in annual payroll savings and a 60% increase in customer service efficiency.

The tools have not triggered layoffs, he said, but they have frozen hiring and freed staff from administrative tasks.

Alicia Hale, svp of growth at B2B SaaS firm Demandbase, said the company runs AI across marketing and sales. The sales team uses AI tools to speed up outreach, which lets the business grow without adding headcount.

"We've been able to keep our hiring flat on the business development side of the house, because instead of having to go and ask for additional resourcing, we can scale the high-quality personnel that we have," said Hale.

On the creative side, the in-house advertising team measures gen AI output by asset volume. Demandbase saved 366 hours in the last 90 days using AI writing tool Jasper for content generation. AI also improved ad efficiency, letting the company reinvest $250,000 back into the platform.

Even so, the precise cost-benefit ratio across Demandbase's overall marketing operation remains unclear. "We've all gotten caught up in this AI frenzy that the diligence isn't being done as well as it could be," said Hale. "It's going to be something that we all have to begin to focus on very quickly."

Zip, a buy now, pay later brand, has applied AI to campaign creation, digital twinning, brainstorming and creative production. Jinal Shah, chief customer officer at Zip, said the biggest measurable difference is the company's ability to scale leadership and employee productivity.

Mastercard has run AI-powered tools for several years. Four years ago it launched a proprietary digital marketing engine that listens to social media, identifies trends, alerts social media staff and automatically selects and launches creative campaigns in real time, according to Guerin. The brand also uses generative AI to recommend, regenerate and edit sonic tracks for campaigns.

Across all cases, calculating ROI in terms of revenue generation or marketing cost savings—the kind that allows apples-to-apples comparisons—has proven harder than the initial setup.

"It's hard to put a value on this right now other than productivity. I have more hours in my day—things like that," Shah said.

Gartner analyst Nicole Greene argued CMOs and C-suite execs may need a different set of expectations, ones not directly tied to financial value.

"Part of the challenge is that AI can't be viewed as a straight technology investment," she said by email. "You can measure that with operational metrics, but in today's volatile world we need to move from productivity to proving growth and financial impact."

In-house slowdown

The measurement gap may be holding back wider deployment. As of last year, most of the industry still had not fully adopted AI in campaigns, according to the Interactive Advertising Bureau's Internet Advertising Revenue Report. Its April report found 70% of agencies, brands and publishers still use AI only in limited ways, citing messy data, security concerns and disconnected tools.

The caution is visible at the top of the market. A year ago, Klarna said AI delivered 37% cost savings of roughly $10 million per year across ideation, image creation and translation. This year, CEO and co-founder Sebastian Siemiatkowski told Bloomberg the company is now prioritizing human support, ensuring customers can always speak to a real person—a flick at the limits of the technology.

Klarna is not alone in its change of heart. Through 2027, Gartner predicts 60% of gen AI projects will be abandoned after proof of concept, driven by inadequate AI-ready data, lagging governance, weak risk controls, steep costs or unclear business value.

The measurement gap also stalls in-housing. CMOs otherwise keen to bring marketing capabilities in-house still depend on agency partners. Demandbase, for example, continues to work with creative and PR agencies, and Hale said there are no current plans to reduce their briefs.

"We really see AI as a way to extend the capabilities of our internal team members," said JJ Kaye, evp and chief marketing officer at Capital Bank. "I'm specifically hiring people now with AI skill sets. [But] for the moment, I don't see it replacing agencies."

In-housing remains one of the industry's key tectonic shifts, and most observers expect AI to accelerate it. But without comparable ROI metrics for AI, few marketers can justify cutting out their agency partner—at least for now. Tech titans, notably Meta, are meanwhile vying for more marketing spend, rolling out AI-powered tools with plans to fully automate AI ads by 2026.

Marketers expect the tools to advance and efficiency to unfold sooner rather than later. Many compare the AI learning curve to that of social media or SEO.

"As a marketer, I will never fall into that debate about something as fundamental as AI," said Zip's Shah. "You fundamentally have to accept that this is changing how we work."