Half of Marketing Teams Are Replacing People With AI. Almost Nobody Can Prove It's Working

A new survey finds 47.8% of marketing organizations swapped human tasks for AI this year. Only 6.5% measure that investment against a benchmark.

Every marketing leader knows the budget meeting where a channel gets cut because nobody could prove it worked. Marketing spent two decades building the tools to survive that meeting: attribution, media mix modeling, incrementality tests.

AI is getting a pass from all of them.

In a survey of 184 marketing and advertising professionals published today by ad-tech company Basis, 47.8% said their organization replaced human tasks with AI in the past 12 months. That's up from 27.9% a year ago and 10.4% in 2023. Only 6.5% said they evaluate their AI investments against defined metrics and benchmarks, according to Basis's fourth annual AI and the Future of Marketing report.

The survey asked about tasks, not jobs, and a replaced task doesn't always mean a lost role. But tasks are what jobs are made of. And most of the teams doing the replacing can't say what they got in return.

The work moved to agents faster than anyone measured it

AI is now a daily habit in marketing. 61.4% of respondents use it every day, up from 33.6% last year and 10.9% in 2023. Only 1.6% don't use it at all.

The bigger shift is who, or what, does the work. 51.1% of marketing organizations now run AI agents, a 155.5% jump in a single year. Those agents handle reporting and analytics (72.3%), draft campaign briefs (56.4%), create content (45.7%) and plan media (35.1%).

That list reads like the job description of a junior marketer. Reporting and first drafts are where people learn the craft, and they're the first tasks to go. When Demandbase and Adobe put agents in charge of campaigns last month, we asked who signs off on the work. The Basis data suggests a harder question: who checks whether it paid off?

Two-thirds can't say whether it's working

65.1% of respondents either don't measure AI ROI at all or don't know whether anyone at their organization does. That's not a rounding error. It's the default.

This isn't new, and marketing isn't alone. Last year, MIT's NANDA initiative found that about 95% of enterprise generative AI pilots stalled with little to no measurable impact on the P&L. The same research found more than half of generative AI budgets going to sales and marketing tools, even though the strongest returns showed up in back-office automation. Marketing is still putting the most money where the payoff is least proven.

The foundation is getting weaker, too. Only 13.0% of marketers describe their first-party data as extensive and well-structured, down from 17.9% a year ago. Agents run on that data. As Bee Enterprises' Siraj Dawar showed us with a 30x ROAS campaign that created almost no new demand, automation scales whatever signal you feed it, good or bad.

Everyone thinks the cuts will happen to someone else

The most human number in the report is a contradiction. 70.0% of marketers see little or no risk that AI will replace their own job in the next three to five years. Yet 50.6% expect substantial job losses across the industry in the same window.

At State of Brand, we think the second number is the honest one. The agencies are already showing it. WPP's headcount fell by more than 10,000 between the end of 2024 and June 2026, and Omnicom moved to cut more than 4,000 roles after closing its Interpublic deal, Communicate reports. WPP has since planned 1,000 more cuts, and one analyst told The Drum they'll likely land at junior levels, where AI is taking over execution work. Omnicom even moved the engineers who built its AI platform to an outside contractor.

Brands aren't immune. HubSpot, the company that taught B2B the inbound playbook, is cutting about 660 jobs as it reorganizes around AI. It's no surprise more marketers are building audiences their employers can't take away.

The money is coming out of people, and the CEO is noticing

AI isn't being paid for with new money. CMOs now put 15.3% of their marketing budgets toward AI, according to Gartner's 2026 CMO Spend Survey, while total marketing budgets sit at 7.8% of revenue, roughly 18% below where they were four years ago. The CMO Survey shows marketing headcount growth down by more than half from last year's rate, with the largest companies shrinking their marketing teams outright. When the budget doesn't grow, AI gets funded by something else, and headcount is the easiest line to move.

That trade, made without a scoreboard, is a gift to anyone who wants to go around the CMO. Publicis is now pitching AI transformation directly to CEOs and skipping the marketing department in more new-business pitches. Gartner warned in February that by 2027, more than 40% of CMOs who push for bigger brand budgets will lose influence with the C-suite because they can't show the return. An AI budget nobody can measure invites the same outcome.

What measuring it actually looks like

It can be done. Lloyds Banking Group ran its marketing AI rollout like an experiment. In a project it called Turing, three ring-fenced teams worked the same real product-launch brief: one with no AI, one augmented by AI, one using only AI. Lloyds says generative AI delivered £50m of value across the group in 2025 and expects more than £100m this year.

Measuring also told Lloyds where to stop. Its top marketer, Suresh Balaji, said at Marketing Week's Festival of Marketing last week that some of the bank's AI content experiments risked degrading the brand. His summary: "AI can raise the floor, but only you can raise the ceiling."

That's the point. A benchmark doesn't just justify AI spend. It tells you which work to hand over and which to keep.

Our take

Hold AI to the standard you hold media to. Every agent gets a baseline, a target and an owner before it ships, the same as a new channel would.

If AI absorbed someone's tasks, say where the time went. Did it become pipeline, speed or a smaller payroll? Those are three different stories, and your CFO will want to know which one you're telling.

Fix the data before you automate on top of it. And write the AI scorecard yourself, before an agency or a consultant writes one for your CEO.

Marketing earned its seat at the table by proving what media spend returns. If it can't do the same for AI, someone else will decide what marketing is worth.

Data: Basis, AI and the Future of Marketing: 2026 Edition, survey of 184 marketing and advertising professionals at agencies and brands, published October 19, 2026. Basis sells advertising technology, including AI tools, and the full methodology sits in its gated report. Gartner spend figures are as reported in secondary coverage of its 2026 CMO Spend Survey.