Revenue per marketer: the only AI benchmark that matters?

Over my twenty years in the marketing industry, activity has generally been constrained by budget, time and resource. In the AI era, that balance has shifted. But have results improved?

Marketing teams are producing more content than ever. A team of twelve can easily generate what used to require thirty. Output is up. Costs per asset are down. The dashboards look excellent.

And yet McKinsey's 2025 Global Survey on AI found that nearly 80% of companies report no significant bottom-line impact from generative AI despite high adoption. Benchmarkit's State of AI in Marketing research shows confidence in proving AI ROI has actually fallen year on year, from 49% to 41%. Across sectors, only 29% of organisations can dependably measure ROI on their AI initiatives.

It’s clear that something is wrong with what we're measuring, not with what we're building.

The productivity trap

The instinct, when you invest in AI, is to measure what AI obviously does. It writes faster. It designs faster. It produces more variations, more drafts, more assets in less time. So teams measure speed, volume, cost per asset. They report that content production is up 300%, or that campaign turnaround dropped from three weeks to four days.

These are real improvements. They are also, on their own, almost completely uninformative about whether the investment is paying off.

Is any of this additional output actually converting? Is it reaching the right people, changing their behaviour, moving pipeline? Research suggests the answer is often no. A 2026 survey by 10Fold Communications found that sixty-seven per cent of marketers now use AI for content creation frequently or all the time, yet MarketingProfs' research from the same year shows only 6% report significant performance improvements. In B2B specifically, industry benchmarking data shows AI-generated content converts to qualified opportunities at 18% lower rates than human-created content. Consumers aren't neutral either: research into consumer sentiment around AI content found that 52% reduce engagement when they even suspect content was produced by AI.

More content that performs worse is not a productivity gain. It's an expensive way to dilute your impact.

A better question

If AI has genuinely improved your marketing operation, the evidence should show up in one place above all others: how much revenue is generated per marketer on your team.

Revenue per marketer is a deliberately blunt metric. It doesn't care how many assets you produced or how much time you saved. It asks one question: did your team generate more revenue with the same people, or fewer?

The logic is straightforward. If AI is making your marketers more effective, each one should be worth more to the business. Better targeting means less waste. Better content means higher engagement. Better audience insight means smarter campaign decisions. These compound.

If AI is only making your marketers faster at producing things that don't convert any better than before, revenue per marketer stays flat. Or it declines, quietly, as you spend more on tools that accelerate the production of material nobody acts on.

The gap is already visible. LinkedIn's Workforce Report shows marketing job postings grew 6% year on year in 2025-2026 while marketing output grew 24%. That sounds like a productivity story. But output growth without corresponding revenue growth is just inflation, and nobody reports inflation as a win.

What actually changes the ratio

Revenue per marketer improves when marketers make better decisions about who to target and what to say to them. Everything else is infrastructure.

Better targeting means understanding your audience segments well enough to concentrate budget where it will actually produce pipeline, rather than spreading it thinly and hoping. Better messaging means producing content that earns attention from a specific reader rather than producing generic material and relying on volume to compensate for relevance.

AI can serve both goals. Most AI being used in marketing doesn't.

The majority of tools in use today are production accelerators. They take a prompt and generate output, optimised for speed. What they don't do is understand why a VP of Procurement in manufacturing responds to risk-reduction framing while a CTO in financial services responds to innovation framing. They write for a statistical average because they have no model of the specific person reading.

Sixty-five per cent of B2B content goes unused, according to Forrester as recently as 2023. That number predates the current wave of AI-generated content but I suspect there's no evidence it's improved.

The difference understanding makes

A useful contrast exists in a less common approach to the problem. Ada Create, a B2B content platform, inverts the sequence most AI tools follow. Where the default workflow starts with generation (write me a blog post, give me an ad), Ada starts with audience understanding. Its Perspective Engine profiles what drives each reader, what motivates their decisions, what creates resistance, what framing they respond to, before the Content Engine writes anything.

In practice, this means content for a CFO in healthcare and content for a CTO in manufacturing aren't surface-level variations of the same piece with a few terms swapped. They're different arguments, built from different evidence, using different framing, because the two readers care about fundamentally different things. Google and Unilever already trust Ada's underlying technology for this kind of audience understanding at scale.

This matters for revenue per marketer because it addresses the waste problem at its source. A marketer using this approach spends their time on strategy and positioning. The AI handles the personalisation that would otherwise require a team producing audience-specific variants by hand, in hours rather than weeks.

That's an AI investment you'd expect to show up in the revenue-per-marketer number. The output is faster, yes. But the reason it moves revenue is that each piece is built for the person reading it.

The question worth asking

Next quarter, when the AI ROI slide comes up, try a different question. Skip the production metrics. Skip the hours saved.

Ask what happened to revenue per marketer.

If the number went up, the investment is working. Your team is generating more revenue per head because AI is helping them make better decisions and produce more effective work.

If it stayed flat despite increased AI spending, you've invested in speed without direction. More content, produced faster, performing no better than what came before.

Most marketing AI investment over the past two years has optimised for the wrong thing. Volume and speed are easy to measure and easy to present to a board. They're also easy to mistake for progress. Revenue per marketer is harder to game, harder to inflate, and harder to ignore when it doesn't move.

Which is exactly why it's the metric worth watching.