Report Your AI Programs in Dollars, Days, and Owners. Coherence at Velocity | October 23, 2026
What will you hand your CFO in January to show that your team should keep its AI budget in 2027?
In my upcoming book Coherence at Velocity I name measurement as one of three enablers of coherence, alongside architecture and governance. Leaders need measurement that separates real improvement from well-documented activity, and most organizations drift back to activity metrics because teams can collect them easily and they make everyone look busy. Campaigns shipped, features launched, and budget spent all leave a clean trail. The question I ask leaders to put to every scorecard is whether the enterprise made the customer experience more coherent while improving its ability to change again.
On Monday I argued that every organization-wide agent needs a registry entry, an owner, and a stopping rule. This edition covers the next question that owner faces: what to report, and to whom. Analysts and authors at Forrester, Gartner, HBR, and MarketingProfs published on it between October 20 and 22, and they agree on the audience. Finance, procurement, and the board now read the AI lines in your plan, and they are asking for outcomes.
Prepare to report AI results in the units your CFO and CEO already use
Mike Proulx and six Forrester colleagues predict that CMOs will scale back a third of their AI initiatives in 2027 as consumption costs rise. David Furlonger of Gartner reports that 44 percent of CEOs struggle to show stakeholders financial returns from AI, and he tells CIOs to measure business outcomes. Justine Hastings, Chad Syverson, and Seth Zimmerman write in HBR that companies deploy AI inside functions and leave the coordination work between functions in place, and they recommend tracking decision quality, cycle time, and operating performance in place of adoption. Mark Boothe, CMO of Domo, writes in MarketingProfs that the CMO still answers for every agent the team runs.
Read together, these authors give you the outline of your 2027 AI report. For each program, show what it costs to run and govern, which business result it moved, and who answers for it.
Expect finance and procurement to review every AI line in your 2027 plan
Mike Proulx, Emily Collins, Dipanjan Chatterjee, Brad Haag, Zeid Khater, Jessica Liu, and Joe Stanhope published Forrester's 2027 predictions for B2C CMOs on October 21. In Forrester's Q3 2026 CMO Pulse Survey, 82 percent of US B2C marketing executives expect to provide stronger justification for future AI investments, and 78 percent say executive leadership demands stronger evidence of value than it did 12 months ago. The authors summarize the change in one line: "AI will get funded because it performs."
Their first prediction is that marketers will scale back a third of AI initiatives because of AI costs. Vendors price by usage, so every prompt, agent interaction, generated asset, and automated workflow adds to the bill, and the authors expect procurement and finance teams to take a larger role in AI decisions as those costs get harder to forecast. CMOs, they write, will cut initiatives whose consumption costs outrun their value and retire applications that agents make redundant. In their second prediction, AI agents replace legacy marketing and insights tools at one in 10 companies. In their third, five brands suffer AI-driven growth setbacks, and 64 percent of the surveyed executives expect their own use of AI to produce a brand-damaging incident within 12 months.
In my AI Capability Framework, compute cost sits among the Forces, the external pressures you contend with, and your choice of mechanism decides how much of that cost each use case carries. The rule I give teams is to use the lightest mechanism that clears the task's accuracy bar. A rules-based send-time model and an agent that reasons through the same decision on every run can reach similar answers at very different prices. When finance asks for the cut Forrester predicts, start with the agentic work a spreadsheet formula could have done.
It reminds me of what Nir Elharar, Chief Marketing Officer at Atera, said when I interviewed him on The Agile Brand podcast: "Your first real agent deployment becomes the internal benchmark for subsequent AI purchases, so pick a use case you'd be willing to be measured against later."
Apply Elharar's advice to reporting as well as purchasing. Your CFO will judge every later AI request against the first program you present with a cost line and a result, so lead with the one you can defend.
My take: Before procurement asks, list every AI program in your 2027 plan with its annual consumption cost and the business metric it moves. Mark each program that has no metric as a candidate for the third you cut.
Report the outcomes your CEO tracks, and drop AI activity from the report
David Furlonger, distinguished VP analyst at Gartner, presented results from the 2027 Gartner CEO Survey at Gartner IT Symposium/Xpo in Orlando on October 20. In the survey of 203 CEOs and senior business executives, fielded in April and May 2026, 44 percent said they struggle to demonstrate AI-driven financial returns to stakeholders. They have kept spending: 75 percent plan to increase technology investment, and 80 percent said technology has become strategy. Sixty-three percent named radically redesigning the operating model as a top move to accelerate AI returns through 2027, and 52 percent named deploying AI agents as employees. Furlonger's diagnosis: "The AI value problem will not be solved by simply adding more AI to existing ways of working."
He gives CIOs three recommendations. Redesign workflows and decisions around AI, build modular architecture so the organization can switch providers and scale new capabilities, and measure outcomes in place of AI activity. He names four outcomes: unit economics, decision velocity, speed to market, and customer outcomes. If your team already tracks a measure like the Decision Latency Index, you have one of the four in hand.
In Marketing Operations 3.0 I define Return on Agent as the value created by agented work, minus the total cost of that work, divided by the total cost. Total cost includes running and governing the work, and most marketing teams undercount that governance cost in their business cases. I also separate driver metrics, which you use to judge whether the system creates more value, from guardrail metrics such as drift rate, compliance exceptions, and brand safety violations, which you use to decide whether the system is safe to scale. Report ROA to your CEO and CFO, and keep the drivers and guardrails as the explanation behind the number.
Furlonger surveyed the executives your CMO reports to, and 44 percent of them told Gartner they struggle to show AI returns to their own stakeholders.
Commentary: Replace every adoption metric in your AI report with one of Furlonger's four outcomes. Put any program your team cannot tie to one of the four on the review list.
Count your coordination costs before you automate, then report the result across functions
Justine Hastings of Auger and the University of Washington, Chad Syverson of Chicago Booth, and Seth Zimmerman of Yale SOM published The Deployment Failure Holding Back Your AI Investments in HBR on October 21. They cite an early-2026 survey of nearly 6,000 senior executives in which 69 percent said their company actively uses AI and about nine in ten reported no measurable effect on productivity or employment. They also cite McKinsey's late-2025 finding that 88 percent of organizations use AI regularly and 39 percent report any enterprise-level EBIT impact.
Their explanation is that companies apply AI inside individual functions and leave the coordination work between functions untouched: pulling data from separate systems, reconciling conflicting information, and aligning teams. They cite research in which employees spend 37 percent of their time on coordination work. Their five recommendations are to measure coordination costs before automating, deploy AI across functions, reward enterprise-wide results, redesign roles as AI takes on coordination, and track decision quality, cycle time, and operating performance in place of adoption.
I have watched marketing teams report a 60 percent drop in drafting time in the same quarter their launch calendar slipped, because the hours they saved went into handoffs with legal, product, and sales operations.
Take a hypothetical B2B software company with $600 million in revenue. Marketing runs 48 product and campaign launches a year, and each launch takes an average of 31 days from brief to live, 19 of them spent in handoffs. The company spends $1.1 million a year on marketing AI.
In the first version, the team buys drafting agents. Drafting falls from six days to two, and the average launch takes 27 days. Two writers per launch save about 3,100 hours a year, worth roughly $230,000 at a $75 loaded hourly rate. Return on Agent comes to ($230,000 minus $1.1 million) divided by $1.1 million, about negative 79 percent. The marketing operations lead walks into the budget review with a strong adoption slide and a negative number.
In the second version, the team times its handoffs first. It then funds a shared library of approved product claims, owned by product marketing and reviewed by legal every month, and an agent that assembles each launch brief from the pricing system, the product roadmap, and that library. The library owner's time and oversight add $150,000, for a total cost of $1.25 million. Handoff days fall from 19 to eight, and the average launch takes 16 days. Coordination labor across marketing, product marketing, and legal falls by about 6,300 hours, worth about $475,000. The drafting savings of $230,000 remain, and with the same staff the team ships 60 launches a year in place of 48. At an average gross margin of $90,000 per launch, the 12 additional launches add $1.08 million. Total value comes to about $1.79 million, and ROA comes to about 43 percent.
The marketing operations lead presents the second version jointly with the heads of product marketing and legal, because part of the savings comes from their teams' hours. In that report the lead can show two of Furlonger's outcomes, speed to market and unit economics, along with the cycle-time measure Hastings, Syverson, and Zimmerman recommend.
It reminds me of what JD Carter, Chief Strategy Officer at Vasion, said when I interviewed him on The Agile Brand podcast: "Have I digitized and centralized all of my data and documents? Once you've done that, you're in an operational place to create interoperability across systems and then implement governance; because of the urgency of AI, many companies are trying to jump and skip these steps and will waste investment."
In the second version, the team follows Carter's order and centralizes the documents before any agent reads them. Agile Brand Principle 2 asks you to know when an evolutionary change will do and when you need a revolutionary one. The second version is the revolutionary change, because the team redesigned how three teams hand work to each other.
My take: Before your next agent purchase, time three recent launches from brief to live and count the days spent in handoffs. Fund the work that removes handoff days before you fund more drafting capacity.
Name the person who answers to the board for each agent your team runs
Mark Boothe, CMO at Domo, published The CMO Managing 3,000 Agents and Zero People Is Going to Fail in MarketingProfs on October 20. He reports that 93 percent of his marketing employees have built their own agents and run active AI workflows, and that his team put more than 20 distinct AI systems into production in 12 months. One non-coder on the events team reclaimed 18 hours a week, and producing keynote visuals in-house saved more than $50,000 in agency cost.
Boothe argues against a model in which one CMO manages thousands of agents with no people. In that model, nobody can see what each agent does, what data it uses, what it costs, why it made a recommendation, or which version is live, and the team finds most mistakes after they cause harm. He states the accountability point directly: "'The model did it' will not survive contact with your board." His team embedded a dedicated AI lead in marketing for a year and built a knowledge graph covering messaging, product, content, brand, and design, so output stays consistent whether a person or a system produces it.
Boothe reports his own results in hours reclaimed and agency fees avoided, which a CFO can check against a timesheet and an invoice.
In Marketing Operations 3.0 I recommend a CO2 Council, a cross-functional governing body that owns the decisions no single function can make. The council uses its dashboard to answer four questions: Is the hybrid workforce producing economic value? Is the system moving fast enough to matter? Are the guardrails holding? Is the customer experience getting better or merely more automated? Each review ends in a decision to scale, fix, reallocate, or kill. The agent owners you named on Monday report to that body, and the council reports to the CEO and CFO.
My take: Give each organization-wide agent owner one line on the council dashboard: the agent's cost this quarter, the business metric it moved, its guardrail status, and the decision the owner recommends.
Take these actions before your 2027 budget locks
List every AI program with its cost and its metric. Include consumption, licenses, and the hours your team spends governing it.
Swap adoption metrics for outcomes. Use unit economics, decision velocity, speed to market, or customer outcomes.
Calculate Return on Agent per program. Report ROA upward and keep driver and guardrail metrics as the explanation.
Time your handoffs. Measure days from brief to live and the share spent in handoffs before you buy more drafting capacity.
Report shared results jointly. When a program saves hours in legal or product marketing, present it with those leaders.
Name an accountable owner. Every organization-wide agent gets a person who answers for it at the council and to the board.
Answer your CFO with a number finance already tracks
In my consulting work with Fortune 1000 marketing organizations, the budget reviews that went fastest were the ones where the marketing lead opened with a number finance already tracked, usually cost per launch or days to revenue. The reviews that ran long opened with usage charts, and the CFO spent the meeting converting them into dollars.
Coherence at Velocity lays out five governance gates for customer-facing change. In Gate E, the post-release review, the team measures whether the change delivered its intended outcome, captures what it learned, and identifies new debt for the next cycle. Most marketing teams I work with run AI programs through intake, design, build, and release, and few hold the fifth review. I read Forrester's 82 percent and Gartner's 44 percent as the cost of skipping it.
Answer the opening question on one page for each AI program in your 2027 plan. Write the business outcome it moved, its total cost to run and govern, its Return on Agent, and the name of its owner. Review that page with your CFO before the budget locks, and send any program without an outcome on the first line back for redesign.
Featured this cycle:
Predictions 2027: CMOs Reckon With An AI Reality Check. Mike Proulx, Emily Collins, Dipanjan Chatterjee, Brad Haag, Zeid Khater, Jessica Liu, and Joe Stanhope, Forrester, October 21, 2026
Gartner Says 44% of CEOs Struggle to Demonstrate AI Financial Returns as Technology Becomes Central to Growth. David Furlonger, Gartner, October 20, 2026
The Deployment Failure Holding Back Your AI Investments. Justine Hastings, Chad Syverson, and Seth Zimmerman, Harvard Business Review, October 21, 2026
The CMO Managing 3,000 Agents and Zero People Is Going to Fail. Mark Boothe, MarketingProfs, October 20, 2026