You Can Prove Everything Except Your AI Spend

Steve Domino ·Partner, Eller Media ·

You Can Prove Everything Except Your AI Spend

You can defend paid media. You can defend the events line, the content line, the headcount line. Then finance points at the row that grew fastest this year, the AI tools and agents, and asks the one question you cannot answer cleanly: what did it return? The uncomfortable truth is that the budget line growing quickest is the one marketing can explain least, and finance has noticed.

This is not a data problem. It is an accountability problem, and it is getting worse while spend goes up.

Key takeaways

  • The ability to demonstrate ROI on AI investments fell from 49% to 41% in a single year, even as AI became the fastest-growing budget line.
  • 85% of marketers say they can measure holistic ROI. Only 32% actually do. Confidence and accountability are not the same thing.
  • Among teams using AI analytics tools, only 29% can quantify the ROI of those very tools.
  • The gap is not missing data. It is that AI activity was never tied to a business outcome the board recognizes.
  • Control Restores Confidence: when the AI line is visible and tied to outcomes, it stops being the row you dread defending.

Why can you prove every line except AI?

Because AI spend grew faster than the accountability around it. The other lines have years of reporting habit behind them. AI arrived fast, got funded fast, and generated activity that no one connected to a business outcome at the budgeting stage. So you can show the tools, the output, and the hours saved, but not the return.

The numbers are moving the wrong way. According to Benchmarkit’s State of AI in Marketing 2026, the share of marketers who could demonstrate ROI on AI investments fell from 49% to 41% in a year. Spend rose, provability dropped. That is the signature of money moving ahead of measurement.

For Skeptical-CFO Sam, this is the whole issue. Sam does not distrust marketing, he distrusts ambiguity. A line he cannot explain to the board is a line he cannot approve, no matter how promising the demo was. When the fastest-growing spend is also the least explainable, Sam stops asking whether AI works and starts asking who signed off on it.

What does the measurement confidence gap actually mean?

It means teams feel measured without being measured. Surveys show 85% of marketers are confident they can measure holistic ROI, while only 32% actually do it. That gap between feeling accountable and being accountable is where budget credibility quietly dies, because the board eventually tests the confidence and finds nothing underneath it.

That 85-versus-32 split, reported in the 2026 attribution data, is the pattern behind a lot of lost budget arguments. Marketing believes it can prove the number. Finance asks for the proof. The proof does not exist in a form finance accepts, and the credibility gap widens.

AI made this sharper, not softer. Among teams using AI analytics tools specifically, only 29% can quantify the ROI of those tools. The instruments meant to measure everything else cannot measure themselves. A brand that assumes new tooling closed its measurement gap usually finds it opened a new one, which is the same lesson behind AI overviews breaking your attribution: more sophisticated inputs do not automatically produce accountable outputs.

What does your CFO actually want from the AI line?

One number, in their language, tied to the business. Not tool adoption, not content volume, not time saved. Finance wants to know what pipeline or revenue the AI line produced and how that compares to its cost. If marketing reports AI in its own metrics, the CFO reads the line as unaccountable and defends the budget accordingly.

This is why so many good AI investments lose their funding. The work may be real, but it is reported in a dialect finance does not speak. We have written before about how your dashboard speaks marketing while your CFO reads money, and the AI line is where that translation gap costs the most, because it is the newest and least trusted spend.

Consider a mid-market team that spent $3,000 a month on AI content tools. Asked for ROI, they showed output volume and hours saved. The CFO cut the line, because none of it connected to pipeline. The tools were producing. Nobody had decided, up front, which revenue number they were producing toward. That decision, not the tooling, was the missing piece.

Why does more AI budget make the problem worse?

Because unaccountable spend does not become accountable by growing. Gartner found 63% of CMOs plan to increase AI spend further, which means the least-provable line is also the one expanding fastest. Without accountability built in first, every added dollar adds activity to explain and widens the gap between what you spend and what you can defend.

More tools also means more places the number can hide. Multi-touch attribution reached 41% enterprise adoption, yet only 18% of those implementations are rated highly accurate by the teams running them, per the Gartner 2026 CMO Spend Survey. Adding AI on top of measurement that is already shaky does not fix the reading. It adds a layer.

This is the same erosion that shows up when finance stops believing in your brand marketing. Belief does not fail all at once. It fails one unanswerable question at a time, and the AI line now supplies the most of them. Growing the line without closing the accountability gap just speeds up the loss of trust.

How do you make the AI line defensible?

Tie it to an outcome before you fund it, then report it in the view finance already trusts. Decide which pipeline or revenue number each AI investment is meant to move, and measure that. ROI you can prove is designed in at the budgeting stage. It cannot be reconstructed from a tool dashboard after the money is spent.

That is what a Scorecard does. Not another dashboard of marketing metrics, but an accountability layer that connects every marketing activity, AI spend included, to business outcomes in one view the board and the CFO both read. When the AI line sits in that view next to the number it moved, it stops being the row you dread and becomes the row you point to first.

Your next step is not a new tool or a new report. Take the single fastest-growing AI line in your budget and write down the one business outcome it is supposed to move. If you cannot name it, you have found why the line is hard to defend, and you have found the first thing to fix. Prove one line that way, and you have the template for all of them.

Frequently Asked Questions

How do you prove ROI on AI marketing spend?

Tie the AI line to outcomes before you fund it, not after. Define which pipeline or revenue number each AI investment is supposed to move, then report that number in the same view finance already trusts. ROI you can prove is designed in at the budgeting stage. It cannot be reconstructed from a tool dashboard later.

Why is AI ROI getting harder to measure, not easier?

Because AI spend grew faster than the accountability around it. The ability to demonstrate AI ROI fell from 49% to 41% in a year as budgets climbed. New tools created new activity that was never connected to a business outcome, so there is more to measure and less of it tied to anything finance recognizes.

What does a CFO actually want to see from marketing AI spend?

One number, in their language, tied to the business. Not tool adoption, not content volume, not time saved. A CFO wants to know what pipeline or revenue the AI line produced and how that compares to its cost. If marketing cannot express the AI line in those terms, the CFO reads it as unaccountable spend.

What is the fix for the marketing measurement gap?

Accountability infrastructure, not another dashboard. A Scorecard that connects marketing activity, including every AI investment, directly to business outcomes in one view finance and the board both trust. The gap is not a lack of data. It is that the data was never tied to the outcomes leadership cares about.

Frequently asked questions

How do you prove ROI on AI marketing spend?
Tie the AI line to outcomes before you fund it, not after. Define which pipeline or revenue number each AI investment is supposed to move, then report that number in the same view finance already trusts. ROI you can prove is designed in at the budgeting stage. It cannot be reconstructed from a tool dashboard later.
Why is AI ROI getting harder to measure, not easier?
Because AI spend grew faster than the accountability around it. The ability to demonstrate AI ROI fell from 49% to 41% in a year as budgets climbed. New tools created new activity that was never connected to a business outcome, so there is more to measure and less of it tied to anything finance recognizes.
What does a CFO actually want to see from marketing AI spend?
One number, in their language, tied to the business. Not tool adoption, not content volume, not time saved. A CFO wants to know what pipeline or revenue the AI line produced and how that compares to its cost. If marketing cannot express the AI line in those terms, the CFO reads it as unaccountable spend.
What is the fix for the marketing measurement gap?
Accountability infrastructure, not another dashboard. A Scorecard that connects marketing activity, including every AI investment, directly to business outcomes in one view finance and the board both trust. The gap is not a lack of data. It is that the data was never tied to the outcomes leadership cares about.