The Measurement Migration Is a Board-Level Decision Now

Three Fridays in a row this month, I have answered the same phone call.

A CEO I have known for years - different person each Friday - calls me on the drive home from a board meeting. The opening line varies. The middle is the same. The board had just asked a question about marketing measurement that the CEO did not know how to answer.

"What is your team's plan for the measurement migration?"

"How do you validate the numbers in the marketing P&L?"

"When the agents we're deploying optimize against your attribution stack, who owns the methodology?"

The CEOs answered the way most CEOs in 2026 are answering questions like these. They nodded thoughtfully. They said the CMO was looking into it. They moved the conversation along. Then they called me on the drive home because they were not sure what they had just been asked.

I have written the last three issues of this letter from the CMO seat and the CFO seat. This issue is from the seat above both of them. The measurement question has escalated to the board faster than most CEOs are acknowledging, and the framing change it implies is the one I most want to put on the record.

For most of the last decade, marketing measurement was a department concern. In 2026 it became a governance concern.

That is a different kind of question. It requires a different kind of answer.

Three structural shifts are doing the work. Each one would have escalated the conversation on its own. The three together have made it unavoidable.

One. Marketing is now a material capital allocation line. When marketing was five percent of OPEX, board oversight was minimal by design. When marketing is twelve to thirty percent of revenue - and at DTC brands at scale it can clear forty - the system that determines how that capital gets allocated is one of the most consequential systems in the company. The board's fiduciary obligation to oversee material capital allocation is not optional. It applies to marketing now the way it has always applied to M&A.

This is the part I think Will Thorndike's The Outsiders gets at better than any book on the subject. Thorndike's argument, in eight portraits of CEOs who outperformed the market by extraordinary margins, is that capital allocation is the CEO's single most important job - and that the discipline most CEOs apply to one part of the P&L (acquisitions, dividends, repurchases) they fail to apply to others. Marketing has been one of the others. The boards now asking these questions are doing what Thorndike's outsiders would have done a decade earlier.

For directors who want the long version of this argument, Michael Mauboussin's writing on capital allocation discipline at Counterpoint Global is the cleanest contemporary treatment of how board-level capital allocation oversight is supposed to work. His Capital Allocation: Results, Analysis, and Assessment is the one I would hand to every new board director.

Two. The board has figured out that the numbers are wrong. Activist investor letters started referencing marketing efficiency with a different kind of precision in 2024 and 2025. Audit committees, separately, started asking pointed questions about how marketing P&L lines were validated. By 2026, the vocabulary incrementality and iROAS shows up in shareholder letters and earnings transcripts with a frequency that did not exist three years ago.

(Rajeev wrote about why calibration is the actual product of measurement in the first issue of The Incrementalist. That argument is the methodology floor under everything I am about to write.)

The board does not need to become statisticians to ask sharp questions. They need to know that platform-reported numbers overstate impact, that traditional MMM gives static snapshots, and that incrementality testing is the calibration anchor. That base layer of literacy is now common in the rooms I am in. Three years ago it wasn't.

Three. AI agents are being deployed against the existing substrate. Every modern company is, this year, deploying AI into a function that allocates significant capital. In marketing that function is, in most companies, still running on attribution numbers that are systematically biased. CEOs are accountable to the board for the systems their company runs on. Deploying an agent into a function whose substrate is wrong is a governance failure that is going to show up in audit-committee conversations in 2027.

This last point is the one I think most boards are still underweighting. Cassie Kozyrkov writes about this category - decision intelligence - better than anyone in the industry. Her central claim, that every decision is the product of two inputs (the question and the substrate the question is asked against), maps directly onto what is happening in marketing functions right now. Her piece Decision Intelligence: A Discipline for Reasoning in the Real World is the cleanest framing of why the substrate is load-bearing in AI-era operations.

I want to be direct about what this means for governance. The board's responsibility for AI deployment includes the data substrate the AI is reading. If the marketing measurement layer is not calibrated, every AI agent deployed against it is, by definition, optimizing against a biased substrate. That is the governance question. It is no longer a marketing implementation detail.

When a board asks about the measurement migration, they are not asking which vendor the marketing team should buy. They are asking a much harder question.

Is this company set up to make capital allocation decisions on real data, or on the data the platforms self-report?

The CEOs I work with who answer this question well share a posture. They treat the migration the way they treat any other systemic shift in how the company makes decisions - not as a procurement event, but as a change in how the leadership team thinks about its own information.

Concretely, this means three commitments at the CEO level.

Commitment one. The CMO and the CFO share a single source of truth. Not a quarterly meeting where they reconcile differences. A shared dashboard with shared definitions, refreshed at least weekly. If marketing reports iROAS to the board and ROAS to itself, the migration has not happened. The version of this commitment I find most credible at the CEO level is the one that puts the CMO and the CFO on a single weekly review with the CEO present at least quarterly. The format matters because the format is what enforces the discipline.

Commitment two. Marketing spend ladders up and down based on causal evidence. This means a CEO must be willing to defend a larger marketing budget when the causal model says brand and mid-funnel are under-invested - which they almost always are. It also means being willing to cut a high-ROAS-looking channel when the causal model shows it is cannibalizing organic. Both moves are politically difficult. Both are essential. The asymmetry I've watched at our customer base: the cuts are easier to defend after the fact than the increases. Boards that watch a CEO institutionalize causal reallocation tend to back the increases the second time.

Commitment three. AI deployments wait for the substrate. A company that deploys AI agents into a marketing function still measuring on platform-reported numbers is funding a system that will allocate budget with confidence in the wrong direction. The right sequence is: migrate the substrate, then deploy the agents. The companies doing it the other way around will discover this in 2027 the expensive way.

If you sit on a board, or you brief a board, here are the three questions for your next quarterly review. These are the questions I would want my own board asking me.

1. What percentage of our reported marketing performance is platform-reported versus independently validated?

If the answer is "all of it is platform-reported," the migration question is overdue. If the answer is "we run incrementality tests on our top channels and calibrate our attribution against them," the company is in the middle of the migration. If the answer is "we operate on a continuous causal measurement system that calibrates across MMM, incrementality, and attribution," the company has completed it.

Most companies cannot answer this question with precision. The fact that they cannot is itself the answer.

2. What is the largest gap we have ever observed between platform-reported and incrementality-validated revenue on a major channel?

If the CMO has never run a test that produces this number, the company has never stress-tested its own data. Most CMOs who have run one rigorous test can name a thirty to seventy percent gap on at least one channel. That gap, multiplied by the channel's spend, is the magnitude of capital that has been allocated against the wrong signal.

This is the question I would ask first if I joined a new board. It cuts through every methodology debate in fifteen seconds. Either the number exists or it does not.

3. When we deploy AI agents into marketing, what is the substrate they reason against?

This is the question almost no board is asking yet. It is the question every board will be asking by the fourth quarter of next year. CEOs who can answer it now will look prescient. CEOs who cannot will look exposed.

This is also the question I think the emerging AI governance frameworks at the SEC, the EU AI Act's high-risk classifications, and the NIST AI Risk Management Framework implicitly require companies to be able to answer at the audit-committee level. If you cannot tell your auditors what data your operating-decision AI agents are reasoning against, you cannot say with a straight face that you have AI risk under control.

The temptation - and the position most CEOs I speak with are in - is to acknowledge the question, agree it is important, and move it to next quarter. I want to be honest about what that delay costs.

The brands that completed this migration in 2024 and 2025 are now eighteen to twenty-four months ahead in a category where compounding matters. They are not just measuring better. They are deciding better. They have rebuilt their channel mix on causal evidence - almost always landing on more brand spend, more mid-funnel, less bottom-funnel duplication. Their CACs have dropped. Their CFOs trust their marketing budgets. Their AI deployments have a real substrate to learn from.

The companies that migrate in 2026 will catch up to that position within a year. The companies that wait until 2027 will inherit a category in which unified causal measurement is the assumed baseline, and they will compete with eighteen months of disadvantage on the most consequential decisions they make as a company. That is the part that compounds.

I am not making the case to panic. I am making the case that the cost of waiting has crossed the cost of moving. That has not been true for most of the last decade. It is true now.

Three things I am watching as this conversation moves into the boardroom.

One. Whether boards add measurement methodology to the audit committee charter. The natural home for this question is the audit committee - the body already responsible for validating financial reporting integrity. Marketing P&L line items are part of financial reporting. The question of whether those numbers are calibrated is, technically, already inside the audit committee's existing mandate. I expect the first board to explicitly add measurement methodology to its charter within the next twelve months, and I expect it to set a pattern.

Two. Whether the next generation of CFO is hired with measurement methodology fluency. I have started seeing CFO job descriptions at high-growth consumer brands that explicitly list "measurement methodology" or "causal inference" as preferred experience. Three years ago that would have been an absurd line in a CFO posting. It is showing up now. By 2027 I expect it to be standard at mid-market and enterprise consumer brands.

Three. Whether activist investors take this argument to the next level. Letters from Elliott, Trian, Pershing Square, and their peers have started referencing marketing efficiency with a sharpness that did not exist eighteen months ago. I am watching for the first activist letter that names measurement methodology specifically as the lever - not just "cut marketing spend" but "the marketing spend is not calibrated, here is the calibrated number we believe is appropriate, here is the holdout result that justifies it." When that letter is written publicly, the category shifts overnight.

I want to close with the caution I owe the reader.

Boards can over-rotate on a new lever as easily as they under-rotate. There are companies whose marketing function is performing fine and whose board is going to hear this conversation and apply pressure that does not fit. Not every brand needs incrementality on every channel every quarter. Some businesses have a clear enough causal picture that the marginal value of additional rigor is low.

The honest framing is this. Boards should be able to answer the three questions above with precision. They should not necessarily treat the answers as a mandate to act before the operating team is ready to do the work properly. A board that asks the question and then demands a six-week implementation is not improving governance. It is creating the conditions for a rushed implementation that will produce numbers nearly as flawed as the ones it replaced.

The boards that handle this well will ask the questions, listen carefully, and then back their CEO in a careful twelve-to-eighteen-month migration that does the work properly. That is the version of board oversight I am rooting for.

If you sit on a board, brief a board, or have just been on the receiving end of one of these questions, I would like to hear how the conversation went. Not to sell. To compare notes on what is actually being asked in the rooms that matter.

If you are a CMO reading this, show this issue to your CEO. The conversation goes faster when you arrive at it together.

Thanks for reading the fourth one.

Tobin Co-Founder & CEO, Lifesight June 2026