Key Takeaways

  • Open Future Forum's new CMO AI Leverage Report benchmarks marketing leaders by their actual decision-making authority over AI spend rather than by title, which is the right cut because the person controlling the budget is rarely the one on the org chart you'd expect.
  • The most useful benchmark inside the report is whether AI spend is producing leverage — more output without more headcount, agency fees, or tool sprawl, rather than the raw percentage of budget going to AI.
  • Cross-referenced against the 2026 CMO Survey (AI at roughly 9.6% of overall marketing budgets) and CMSWire's read on flat 7.7% marketing budgets, the story is that AI dollars are growing inside a budget that isn't, which means every AI dollar has to displace something.
  • If your AI budget is going to seats and pilots rather than skills, workflows, and repos your team actually shares, you are almost certainly behind the leverage curve regardless of what you're spending.
  • The right benchmark question for a CMO in 2026 is what a dollar of AI spend returns in output per person on the team.

A benchmark report landed this month that I think most marketing leaders should read twice, and the second time with a highlighter. Open Future Forum surveyed marketing leaders who actually control AI budgets, then segmented the respondents by decision-making authority over marketing AI investment rather than by title. That distinction sounds academic until you sit in the room where the AI budget gets decided and realize the person signing the PO is often not the CMO. Sometimes it's a VP of Growth, sometimes it's a founder, sometimes it's a marketing ops lead who quietly owns the stack.

What the report gives you is the first honest peer view of how that money is actually moving. Set it next to the 2026 CMO Survey and CMSWire's piece on flat budgets, and a fairly specific picture forms about who is getting leverage from AI and who is just paying for it.

What the Leverage Report actually measures

The Open Future Forum report does something I haven't seen done cleanly before: it defines respondents by their control over the AI budget, not by title. So a VP of Marketing at a 400-person SaaS company who signs off on the AI stack counts, and a CMO at a larger company whose CIO has quietly annexed all AI spend does not. The interesting question isn't what CMOs think about AI, it's what the people writing the checks are actually buying.

The underlying idea is a Leverage Index: whether AI is helping marketing teams grow output without growing headcount, agency spend, or tool sprawl. That framing matters. The two most common benchmark questions, how much are you spending, and on what, give you numbers that make you feel either ahead or behind, and neither feeling is actionable. Leverage is actionable. Leverage is a ratio, and ratios tell you whether to keep going or stop.

Set it next to the other numbers on the table

The Duke CMO Survey's Highlights and Insights Report puts AI at roughly 9.6% of overall marketing budgets and 9.0% of revenues, and calls out plainly that adoption is outpacing organizational readiness. That last phrase is the one to circle. It means the spend is happening faster than the operating model can absorb it, which is exactly what you'd expect if leaders are buying seats and pilots without redesigning how the work gets done.

CMSWire's read on the same period is that marketing budgets flatlined at 7.7% of company revenue, with 39% of CMOs planning both agency reductions and labor cuts. So the AI line is growing inside a budget that isn't. Every AI dollar has to come from somewhere, and it's coming from agencies and headcount. Which is fine if the AI is producing more output than the agency or the headcount did, and a disaster if it isn't.

Lemniscate has a narrower slice on AI search allocation running 8 to 15% of combined search and content budgets in 2026, up from under 3% two years ago. That's a real reallocation happening quietly under the covers of what still looks like a normal search line item.

Put those together and the honest read is this: the average marketing leader is spending roughly a tenth of their marketing budget on AI, that budget isn't growing, and it's being paid for by cutting agencies and people. The Open Future Forum benchmark is useful because it lets you check whether the tenth you're spending is actually returning leverage, or whether you've just moved money from one line to another and hoped.

The move underneath the benchmark: skills, not seats

Here's what I keep noticing in our own work and in conversations with marketing leaders. The teams getting real leverage from AI have stopped buying AI as seats and started building AI as skills their team shares.

What I mean by that concretely: a seat is a Copilot license or a ChatGPT Teams subscription, everyone has it, nobody's workflow has actually changed, and the invoice grows every quarter. A skill is a specific, repeatable capability that lives in a repo, gets version-controlled, and improves every time someone uses it. A prompt engineering skill, a brand voice skill, a component build skill, a research synthesis skill. The teams doing this well keep their skills in something like GitHub, versioned, so anyone can pull down the same sharpened tool the person next to them used yesterday.

The reason this matters for a benchmark report is that dollars-spent is a terrible proxy for value-received when the underlying technology is this new. If you and I both spend 10% of our marketing budget on AI, but your team is running shared skills through Claude Code with a design system, tokens, and page build rules that produce a live component in twenty minutes, and my team is generating a lot of first-draft blog posts, we are not in the same benchmark bucket. We just look like we are.

That's the gap the Open Future Forum report is trying to name with the Leverage Index framing. Read the report less as "am I spending the right amount" and more as "am I in the leverage cohort or the seat cohort."

How I'd use this benchmark next week

If I were sitting down with the Open Future Forum benchmark and my own 2026 plan, I'd do three things in order.

First, I'd map every dollar of my current AI spend to one of two columns: seats or skills. Seats are recurring per-user licenses with no shared artifact. Skills are shared, versioned, reusable capabilities. This is uncomfortable, because most teams find that 70 or 80% of their spend is in the seats column and they can't point to what changed.

Second, I'd pick one workflow, content production, campaign briefs, competitive research, landing page builds, pick one, and rebuild it as a shared skill with a real repo behind it. Not a Notion doc. A repo. That's the practice that gets you from "we use AI" to "our team uses the same AI capability the same way, " which is where leverage actually starts. If you want a walk-through of how to structure these, our AI marketing blueprint lays out the setup.

Third, I'd set my own leverage metric before I look at anyone else's benchmark. Output per person on the team, month over month. Campaigns shipped per quarter with the same headcount. Pages built per designer. Whatever the honest unit of output is for your team. If that number isn't moving, no amount of AI budget is buying you anything the benchmark would count.

When this whole exercise is the wrong call

Honest limitation, because the benchmark won't tell you this. If your marketing org is under 8 people and the work is mostly custom strategy, brand, and executive relationships, chasing an AI leverage benchmark is probably a distraction. The leverage math works when there's a repeatable output volume to compress. It doesn't work when the work is bespoke by design.

Same goes for hiring an agency to help you build this out, including us. If you don't have a specific workflow you want to compress and a specific output metric you want to move, an agency engagement will produce a lot of interesting artifacts and no leverage. The right move in that case is to run the seats-vs-skills audit yourself for a quarter, find the one workflow that's begging to be rebuilt, and then bring in help.

The benchmark is a mirror, not a map. What I'd do Monday morning is open my own AI line items, sort them into seats and skills, and see which column my last quarter of invoices actually lived in. That's the number worth staring at first.

Frequently Asked Questions

What percentage of a marketing budget should go to AI in 2026?

The current benchmarks put AI at roughly 9 to 10% of overall marketing budgets, based on the 2026 CMO Survey, with narrower slices like AI search running 8 to 15% of combined search and content budgets. But the more useful question is what your AI spend is returning in output per person, because two teams spending the same percentage can be in wildly different leverage positions.

Who should actually control the AI budget inside a marketing org?

The Open Future Forum benchmark segments respondents by real decision-making authority rather than title, and the honest answer is that it varies. In practice it's whoever owns both the workflow being changed and the tool stack enabling it. That's often the CMO or VP of Marketing, sometimes a marketing ops lead, and increasingly a founder in smaller companies.

How do I know if my AI spend is producing leverage or just cost?

Pick an output metric, pages shipped, campaigns launched, briefs completed, whatever is real for your team, and track it per person per month against your AI spend. If the ratio is improving, you have leverage. If spend is up and output-per-person is flat, you've bought seats, not skills.

Is the CMO AI Leverage Report free to access?

The report is published by Open Future Forum. Access details are on the Open Future Forum research page linked in the Sources below.

Sources