What CMOs should fund in AI this year - and what to cut
Marketing budgets now put 15.3% into AI. Only 30% of teams can actually scale it. That gap is the whole problem, and it is a sequencing problem. Most CMOs are funding tools before they have built the data and governance to use them.
The fix is order: foundations first, then agents that close a loop, then scale. Here is the fund, cut, and defer call.
Short answer: CMOs who cannot scale AI are spending ahead of their readiness. They allocate 15.3% of budget to AI while only 30% of teams have the data and governance to use it well. Fix the order. Fund agents that close a measurable loop, cut one-shot content tools that never learn, and defer anything that depends on data infrastructure you have not built.
How much of the marketing budget should a CMO allocate to AI in 2026?
There is no right number. The right ceiling is whatever your readiness can absorb, and for most teams that sits below what they already spend.
The headline figure looks decisive. Gartner's 2026 CMO Spend Survey polled 401 CMOs between January and March and found them allocating 15.3% of budget to AI. The number underneath it is the one that should set policy. Only 30% of those teams rate their own AI readiness as mature. The other 70% are funding a capability their data and governance cannot yet carry.
Readiness leaders spend more, and they have earned the room to. The same survey found AI-ready organizations putting 21.3% of budget into AI. That is the sequence working: foundations first, then a share scaled to match. Read the two figures the wrong way and you conclude you are behind on the percentage. Read them properly and the percentage follows the foundation every time.
So the allocation question has the wrong subject. Before you ask what share of budget goes to AI, ask what share of your current AI spend closes a loop you can measure. For most teams the honest answer is small. That answer is your real ceiling for the year.
What is the CMO AI readiness gap, and why does it decide your budget?
The readiness gap is the distance between AI spend and the data, governance, and skills that make the spend pay. It decides your budget because the foundation is what separates the returns from the write-offs. The tool never does.
Gartner put a multiple on it. Organizations with successful AI initiatives invest up to four times more, as a share of revenue, in data quality, governance, AI-ready people, and change management than the ones with poor AI outcomes. That investment, not the model, is where the returns come from. In the same work, just 39% of technology leaders said they were confident their AI investments would pay off financially, which tells you how rare that ground still is.
The marketing version is starker. BCG's 2026 work on agentic marketing found that only about a third of the CMOs who claim AI is driving transformation have done the foundational work behind it. The rest are reporting transformation they never built. That is the readiness gap in one sentence, and it is why a budget review that opens with tools opens in the wrong place.
Here is the test. Pick any AI line item. Trace where its inputs come from and where its outputs go. Clean, owned inputs feeding a decision someone acts on means the item sits on real ground. Stale exports feeding a slide nobody opens means no model will save it. You are funding motion that returns nothing.
What AI investments are actually delivering ROI for marketing teams?
Agentic AI that closes a measurable loop. The work that pays runs a task end to end, checks its own result, and feeds back a number you can defend to a CFO.
The upside is real and it is sized. BCG reports that agentic AI can deliver 5 to 10% incremental top-line growth and 15 to 20% cost efficiencies. Early adopters have tripled their ROI and cut campaign cycle time. McKinsey's work on the agentic advertising economy finds agentic AI could power as much as two-thirds of current marketing activities, and that more than 90% of advertisers already use AI to plan media, set budgets, optimize targeting, and generate creative. Capability is no longer the constraint.
Use is climbing to match. The CMO Survey, run by Duke's Fuqua School with Deloitte and the AMA across 308 senior marketing leaders, found AI's share of marketing activities nearly doubling in two years, from 13.1% in 2024 to 24.2% in 2026. Generative AI use grew 220% over the same stretch. Those leaders project AI will run 55.9% of marketing activities inside three years. The direction is set.
The loop is what separates ROI from noise. An agent that watches campaign performance and shifts spend toward what is working closes a loop, because the result it produces becomes the input it learns from next. It compounds. A tool that drafts a thousand social posts and stops is a faster way to make the same thing once. Fund the agent that learns.
Doing your own budget review this quarter? Download the human-writing skill. It is the method we use to keep analysis like this from reading like every other AI memo.
What marketing AI investments should you cut?
Cut the tools that produce output without learning from it, and cut the martech you pay for but barely use. Both look like progress on a budget line and return nothing you can trace.
Start with utilization, because it hides in your renewals. Scott Brinker's Martech for 2026 report found large organizations owning an average of 10.2 martech tools while actively relying on four or five. Martech as a share of marketing budget hit a five-year low of 19.4% this year, down from 26.6% in 2021, and the report's Gartner-sourced utilization figures sit in the low thirties. Hold those numbers against your own stack. Half of what you license is idle, and the consolidation trend says the sharp teams have already cut it.
Then cut the one-shot content tools. A generative platform that turns one brief into fifty variations earns its keep once per brief and then waits. It never sees how the variations performed, so it never improves at the only thing that matters. You are paying a subscription for a faster photocopier. Output volume rises, the return stays flat, and eventually the CFO asks why.
The hardest cut is the agency line for "AI-assisted" work you could now run yourself. When an agency charges a premium to operate a model your own team can operate, the premium buys a capability that is no longer scarce. Keep agencies for the strategy and judgment AI cannot do. Stop paying them a markup on the part it can.
What should a CMO defer rather than cut?
Defer the AI that needs infrastructure you have not built. Deferring means you decline to fund a tool into a foundation that cannot hold it, and you wait. That discipline keeps you out of the cancellation statistics.
The cancellations are coming, and they are predictable. Gartner expects more than 40% of agentic AI projects to be cancelled by the end of 2027. The firm names escalating costs, unclear business value, and weak risk controls. None of those causes is a weak model. Each is a project funded before the ground was ready. Deferring is how you stay out of that 40%.
There is a vendor trap to defer around too. Gartner estimates that of the thousands of vendors marketing themselves as agentic, only about 130 are genuine agents. The firm calls the rest "agent washing." A vendor who cannot show you the loop, where the agent acts, checks, and adjusts on its own, is selling an assistant at an agent's price. Defer it until they can show the loop, or until you have built the data layer a real agent would need.
How do you turn this into a budget decision?
Run every AI line item through one screen. Does it close a measurable loop, and is the foundation under it real? The answer sorts each item into fund, cut, or defer.
The table below is that screen. Take it into the review with your current line items and place each one.
| Line item | Decision | The test it has to pass |
|---|---|---|
| Agentic AI that monitors and reallocates spend or optimizes campaigns in a loop | Fund | Produces a measurable result that becomes its own next input, on data you own |
| Foundation work: data quality, governance, AI-ready skills | Fund | The four-times-more spend that separates winners from write-offs |
| One-shot content generators that never see results | Cut | Output volume rises, the return stays flat, nothing learns |
| Martech you license but a handful of people actually use | Cut | Utilization sits in the low thirties; the stack is bloated |
| Agency markup on model work your team can now run | Cut | You are paying a premium for a capability that is no longer scarce |
| "Agentic" tools that cannot show you the loop | Defer | Likely agent washing; wait for the loop or build the data layer |
| AI that depends on data infrastructure you have not built | Defer | Funding it now puts you in the 40% that gets cancelled |
The pressure to spend will not let up. The 2026 CMO Spend Survey found marketing budgets reaching 7.8% of company revenue, 70% of CMOs naming AI leadership a critical goal, and 56% saying they lack the budget to deliver their 2026 strategy at all. You do not win that squeeze by raising the AI share. You win it by moving dollars off the items that cannot close a loop and onto the ones that can. Sequence the foundation. Fund the loop. Let the percentage follow.
Frequently asked questions
How much of the marketing budget should a CMO allocate to AI in 2026? Match the allocation to your readiness rather than to a benchmark. Gartner found CMOs averaging 15.3% of budget on AI while only 30% of teams can scale it, and AI-ready organizations spend 21.3%. The percentage should follow the foundation. Fund what your data and governance can support, then scale the share as readiness grows.
What AI tools are actually delivering ROI for marketing teams? Agentic AI that closes a measurable loop. BCG reports agentic AI delivering 5 to 10% top-line growth and 15 to 20% cost efficiencies, with early adopters tripling ROI. The work that pays runs a task end to end, checks its own result, and feeds back a number you can act on. Tools that generate output and never see how it performed do not compound.
What is the CMO AI readiness gap and how do you close it? It is the distance between AI spend and the data, governance, and skills that make the spend pay. Gartner found successful organizations invest up to four times more in those foundations. Close it by funding data quality and governance before tools, and by tracing every line item to a loop it can measurably close.
Should CMOs cut agency spend to fund AI? Cut the markup and keep the partnership. When an agency charges a premium to operate a model your own team can now run, that premium buys a capability that is no longer scarce. Keep agencies for the strategy and judgment AI cannot do. Move the model-operation dollars in-house and point them at agentic work that closes a loop.
What marketing AI investments are failing and why? The ones funded before the foundation was ready. Gartner expects over 40% of agentic AI projects to be cancelled by the end of 2027 on escalating cost, unclear value, and weak controls. One-shot content tools fail quietly: output volume rises while the return stays flat, because nothing in the loop learns.
What to take into your budget review
Pull your AI line items and run each through one screen: does it close a measurable loop, and is the data under it real? Fund the agents that learn. Cut the generators that do not and the martech nobody uses. Defer anything that needs a foundation you have not built. The percentage is not the decision. The sequence is.
We write about what it takes to run AI-native sales and marketing, down to the budget calls a CMO has to defend in a board review. Download the human-writing skill. It is the free method we use to keep writing like this from sounding like AI, and it is the same skill behind every piece here.
Sources
- Gartner, "2026 CMO Spend Survey Finds CMOs Allocate 15.3% of Marketing Budgets to AI, But Only 30% Are Ready to Scale AI Capabilities" (May 11, 2026). gartner.com
- Gartner, "Organizations With Successful AI Initiatives Invest Up to Four Times More in Data and Analytics Foundations" (April 16, 2026). gartner.com
- Gartner, "Over 40% of Agentic AI Projects Will Be Canceled by End of 2027" (June 25, 2025). gartner.com
- Gartner, "40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026, Up From Less Than 5% in 2025" (August 26, 2025). gartner.com
- BCG, "Making the Agentic Marketing Transformation a Reality" (2026). bcg.com
- McKinsey, "The Agentic Advertising Economy: From Attention to Action" (2026). mckinsey.com
- The CMO Survey, Duke Fuqua School of Business with Deloitte and the AMA, "CMOs Face Headwinds Even as Marketing Value and AI Impact Grow" (January 2026, 308 senior marketing leaders). fuqua.duke.edu
- Scott Brinker, "Martech for 2026 Report" (December 2025). chiefmartec.com
By Christopher Kliebenstein. We build and run AI-native sales and marketing for operators who are shipping it to production.