In one year, the percentage of CEOs who classify marketing as a cost center moved from 35% to 60%. That is a 25-point swing in a single survey cycle. The Boathouse 2026 study (N=150 US CEOs across 16 industries) is the data point, but the data point is not the news. The news is what the swing implies about the seat itself.
The CMO job description has been re-written. The org chart has not caught up.
CEOs now hold marketing to engineering-grade accountability — revenue impact, financial trace, ROI on AI spend — and they hold it 4× harder for CMOs than for any other executive in the C-suite, according to the same study. Meanwhile, marketing teams continue to operate without the engineering-grade infrastructure that would make that accountability possible. No audit trails on what the AI tooling did. No quality gates on what the brand-side agents shipped. No observability on which channel, which campaign, which prompt produced which dollar.
You cannot meet engineering-grade accountability without engineering-grade infrastructure. The labor market already knows this. The org chart is the last thing to find out.
The Two Numbers That Define the Seat
Two data points, separated by industry, tell the same story.
The first is from Boathouse, surfaced through SmartBrief: 60% of CEOs classify marketing as a cost center, up from 35% the year before. The interpretive line from Sonia Chung, the firm’s Chief Strategy Officer, is the one to keep: “CEOs want revenue impact and financial accountability while many CMOs still report in marketing language.” Translation: the language gap is the credibility gap. CEOs are reading P&L; CMOs are presenting reach.
The same study found that CMOs face 4× more AI ROI accountability than other executives. Read that twice. Not “as much as.” Four times. Marketing was the first function where AI tooling spread broadly — content generation, ad optimization, audience modeling, agent-driven outreach — and it is the first function where the bill came due.
The second number is from Kevin Indig’s Growth Memo, analyzing 946 SEO and growth-marketing roles posted between December 2025 and March 2026. The median salary premium for “AI” in the job title is 27%: $113,625 versus $89,438. That is the labor market’s vote on what the work is worth.
But the more useful number is buried one layer deeper. Only 15.5% of postings list AI in the title. 59.5% list AI in the description. Filter by AI in the title and you miss 80% of the AI-required roles. At the director and executive tier, the premium widens to roughly $35,250 and 78.3% of those listings mention AI in the description. For roles requiring 9+ years of experience, 92% mention AI; for entry-level (0–1 year), only 40.9%.
The signal is sharp: the work is already AI-native at the senior end and labeled AI-native at the junior end. The middle is invisible. The labor market has repriced the entire function and the recruiting filters have not kept up.
Engineering-Grade Accountability, Marketing-Grade Infrastructure
The structural problem is the gap between what is being measured and what can be observed.
CEOs ask: which AI investment produced which dollar of revenue? In an engineering org, you answer that with a deployment ID, a feature flag, a request trace, a revenue event, and a join key. The answer is not perfect, but it is auditable. You can show your work.
In marketing, the same question hits a wall. The ad platform that ran the campaign holds the conversion attribution. The CDP that segmented the audience holds a different version. The agent that drafted the email holds none of the conversion data. The CMS that hosted the landing page may or may not have logged the prompt that generated the headline. The CMO is asked to attribute a number whose chain of evidence lives in five vendors and one Google Sheet.
We have written about this in the attribution loop you cannot audit. The brand-side agent that personalized an email two weeks ago cannot tell you why. The reasoning chain is gone. The prompt is gone. Vendor APIs return the rendered output, not the route the model took to produce it. Engineering-grade ROI accountability requires engineering-grade observability. Marketing has been issued the first without the second.
This is not a CMO failure. It is a tooling failure that is being scored as a CMO failure.
The Labor Market Sees the Two-Tier Function
The 27% premium is not a curiosity. It is the market separating two species of marketer.
Tier one is the marketer who can operate brand-side AI agents the same way an SRE operates a production service. They can describe their stack. They can audit their prompts. They can show what got shipped, what got pulled back, and why. Josh Peacock, cited in the Growth Memo analysis, names what is being paid for: “scalable systems built with AI” combined with “precision judgement.” The premium is paid for the combination, not the keyword.
Tier two is the marketer who runs AI tools the way the function has always run tools — through the vendor’s UI, accepting the vendor’s metrics, reporting reach and engagement to a CEO who is reading it as cost.
The compensation gap (+27% median, +$35,250 for directors) is the market’s bet that tier one will be the only tier in three years. The 92% AI-mention rate at 9+ years of experience is the same bet, paid in advance. By the time most marketing orgs finish writing AEO into the strategy deck, the senior labor pool has already been re-graded.
The 80% invisible-AI problem makes this worse for the function. A CMO trying to hire tier one through standard recruiting filters — search “AI marketing director” — will see 20% of the available pool. The other 80% are in postings where AI lives in the description but not the title, because hiring managers wrote the post the way they have always written posts. The filter that matters is not “AI in title” but “AI shaping the work.” That filter does not exist on most ATS systems yet.
What the Seat Actually Is Now
The CMO seat in 2026 is the engineering-leadership seat marketing did not ask for and is not yet structured to fill.
Look at the four floors of accountability the seat now carries:
Revenue trace. Every dollar attributed to marketing must survive a CFO’s audit. The 60%-cost-center reclassification is what happens when revenue trace fails. The fix is not better reporting; it is auditable infrastructure underneath the reporting.
AI ROI defense. 4× more AI ROI exposure than any peer means the CMO is the first executive who has to answer for autonomous spend. When a brand-side agent writes 10,000 personalized emails this quarter, the CMO is the one who can be asked, line by line, what each one cost and what each one returned. The CTO is asked about systems. The CMO is asked about decisions the systems made.
Talent two-tier. The +27% premium is a permanent two-tier compensation structure inside one job family. CMOs need a pay band that reflects this, a hiring funnel that finds the 80% invisible AI-fluent population, and a development path that moves tier-two people into tier one. None of this is solved by HR’s standard career framework, which assumes the job is one job.
Vendor governance. The five-vendor attribution chain is a procurement and integration problem that lives in marketing. Engineering would call this “platform strategy.” Marketing has historically called it “the agency relationship.” The seat now requires the first language, not the second.
These four floors are the same shape as the floors engineering shipped a decade ago. We argued in the agent containment stack post that engineering did the floors in a specific order: compute, data, knowledge, identity. Marketing has to do them too — in a different order, on a different timeline, but as floors, not as quarterly objectives.
The CMO who builds the floors becomes the function’s CTO equivalent. The CMO who reports reach to a CEO reading P&L stays in the cost-center column.
The Move That Closes the Gap
There is no clever fix. There is one structural move.
Stop running the marketing function as a campaigns-and-channels operation with AI tools attached. Start running it as a measured, auditable, observable system that happens to do brand work. The shift is the same shift engineering made between 2010 and 2020 when it stopped being a code-shipping operation and became a platform operation. The CTO seat changed because the work changed. The CMO seat is in the same year-zero moment.
Concretely, three things move:
The reporting language moves from reach and engagement to revenue events with audit trails. Not because reach and engagement are wrong, but because they are not what is being asked. The CEO is reading the same dashboard the CFO reads. The CMO who reports in that language stops being a cost center on the slide.
The hiring filter moves from job titles to evidence in resumes. As the Growth Memo recommends to candidates: AI evidence in the top third of the resume. The hiring side has the dual move: read the description, not the title; formalize the two-tier pay band; and run the funnel against the full 100% pool, not the 20% that says “AI” first.
The infrastructure moves from vendor dashboards to a marketing observability layer the CMO controls. We argued in marketing machine-readability as a KPI that the surface area marketing produces is now consumed by agents as much as by humans. The implication is the same here: the function needs the platform, not just the campaign.
CEOs are not wrong to hold marketing to engineering-grade accountability. AI made that accountability possible and unavoidable. What CEOs are wrong about is assuming the function already has the infrastructure to meet it. It does not. The CMOs who build it become the next CTOs. The ones who do not stay in the 60%.
The seat was rewritten. The chair fits the engineer.
This analysis synthesizes How CEOs View CMOs (SmartBrief, April 2026, citing Boathouse 2026 study, N=150) and The AI Skills Salary Premium (Growth Memo, April 2026, N=946 SEO roles).
Victorino Group helps marketing leaders adopt the observability and accountability infrastructure that AI ROI now demands. Let’s talk.
All articles on The Thinking Wire are written with the assistance of Anthropic's Opus LLM. Each piece goes through multi-agent research to verify facts and surface contradictions, followed by human review and approval before publication. If you find any inaccurate information or wish to contact our editorial team, please reach out at editorial@victorinollc.com . About The Thinking Wire →
If this resonates, let's talk
We help companies implement AI without losing control.
Schedule a Conversation