What would you stop funding in 2027 if every agency, AI tool, and service team billed you only for what customers received?
In my upcoming book, Coherence at Velocity (De Gruyter Brill, 2027), I describe the North Star as a standard strict enough for leaders to tell enterprise capability apart from enterprise theater. The manuscript lists the usual forms of theater: platforms implemented faster than the organization can absorb them, new scorecards and launch rituals sitting on top of the same unofficial spreadsheets, and victory declared because a team shipped something while customers meet the same contradictions with better automation around them.
Between November 16 and November 18, analysts at Gartner and Forrester and a senior partner at McKinsey published four pieces that apply that standard to three budget lines most CMOs will sign in the next 60 days: agency fees, AI tools, and the teams that keep customers after the sale.
Recognize that your contracts and scorecards still pay for motion
Jay Wilson of Gartner predicts that agency fees will drop from 24.5 percent of marketing budgets at the start of the decade to 15 percent by 2030, and he tells marketers to stop paying agencies for activity. Michael O'Grady, Brian Hopkins, Kevin Ogunsua, and Greg Zorella of Forrester argue that minutes saved fail as proof that an AI investment creates value. Shari Srebnick of Forrester tells customer success leaders to measure customer progress before they automate anything. Brooke Weddle of McKinsey says AI transformation starts with redesigning workflows, roles, and skills.
Read together, the four authors describe one leadership job. Each budget line in your plan carries a definition of success, and in most marketing organizations that definition still counts hours billed, minutes saved, or accounts touched. Rewrite the definition before you renew the spend, because your agencies, your vendors, and your own teams will optimize for whatever you agree to count.
Reprice your agency relationships so productivity pays both sides
Wilson, VP Analyst in the Gartner Marketing practice, published Gartner Predicts That By 2030, Agency Fees Will Represent 15% of Marketing Budgets on November 16. He names three drivers: AI, insourcing, and financial pressure on CMOs. Gartner frames the figure as a share of total marketing budgets, which leaves room for absolute agency spending to hold steady, and Wilson says marketers are moving efficiency gains out of research, content, and analytics and into media. He expects more fixed-fee, project-based, and outcome-based pricing, along with hybrid models that pair a base fee with performance components. "Agencies should not be penalized for becoming more productive," he says. "The key is to move from paying for activity to paying for clearly defined value."
Wilson's first line describes a problem that hourly billing creates on its own. An agency that uses AI to cut production time by a third earns a third less under a time-and-materials retainer, so the contract discourages the very productivity the client wants. Marketing Operations 3.0 treats marketing, sales, service, and product as one Commercial System, and a pricing model belongs in that system design. Write the agency's incentives into the same outcomes your internal teams carry, or expect the two groups to pull in different directions.
It reminds me of what Matt Kelly, Growth Strategy Partner from Mavlers, said when I interviewed him on The Agile Brand podcast: "I would say a modern brand-agency relationship that I feel is worth investing in and thinking about and looking for is an agency that is making your own team better."
My take: Wilson's 15 percent figure will tempt procurement to treat this as a cost-cutting exercise. Use the renewal to name two or three outcomes the agency shares with you, and add one more clause covering what the agency will teach your in-house team this year.
Measure AI by the decisions and customer outcomes it improves
O'Grady, a forecast analyst, Hopkins, VP of the emerging tech portfolio, and principal analysts Ogunsua and Zorella published AI ROI: Move From Minutes Saved To Value Created on November 18, drawing on Forrester's report Artificial Intelligence ROI: Productivity Plus Adaptive Advantage. Their first finding: only about a fifth of US enterprises with 250 or more employees that use AI are building new workflows around it. "Adding an assistant to an unchanged process is not enough," they write. Their second finding sets a high bar for productivity alone. Combined estimates put AI's productivity uplift at 1.5 to 2.6 percent of US GDP by 2035, and the authors treat those figures as expectations to be earned. Their third finding repeats a number from Zorella's earlier work: models account for only 10 to 25 percent of total AI spend. The bigger opportunity, in their account, is adaptive advantage, meaning better decisions, faster experimentation, revenue growth, resilience, and quicker responses to customers. "Minutes saved make a persuasive demonstration but do not, on their own, prove that an investment creates value," they write. They recommend managing AI as a portfolio of value, tracking decision quality, experimentation, and customer outcomes next to output per worker, and governing the decisions across functions.
My AI Capability Framework puts goals and strategy at the apex for this reason. The Insights pillar describes the shift the Forrester team is asking for: understanding has become cheap, and decisions remain scarce. A return on investment calculation built on hours saved counts the cheap part and leaves out the scarce part. Most AI business cases I review still lead with the hours figure, usually because it was the easiest number to collect in the pilot.
My take: For every AI tool up for renewal, ask the owner to name one decision the tool improved and one customer outcome that moved. If the owner can only report hours, fund the tool for another quarter and make the decision metric the renewal condition.
Give customer success the full context, then judge it on customer progress
Srebnick, a principal analyst at Forrester, published AI Is Changing Customer Success But Not Its Mission on November 18. She draws on four industry events since September, Bain's "The Rebirth of Customer Success," and Gainsight's restructuring, which its CEO describes as "refounding" the company around the agentic economy. She describes four shifts: customer context available to humans and AI alike, outcomes weighted above output, postsale roles redesigned around the work, and more weight on human judgment as execution gets cheaper. "AI without context can be worse than no AI at all," she writes. Her guidance on measurement: "Don't automate activity for activity's sake. Measure what helps customers make progress, and then automate what matters."
Srebnick's context point is the Golden Record argument from Marketing Operations 3.0: humans and agents need the same unified customer profile, and an assistant that summarizes an account without its support history will sound confident and miss the renewal risk. The fifth of the Agile Brand principles asks brands to focus on the relationship over the transaction. In a subscription business, net revenue retention shows whether the relationship is holding, and a count of check-in calls logged does not.
It reminds me of what Rachel Orston, Chief Customer Officer from Instructure, said when I interviewed her on The Agile Brand podcast: "The strongest measure of customer experience is whether customers achieve meaningful outcomes and continue to see value over time."
My take: Before you deploy AI to your customer success team, list the data sources a senior CSM checks before a renewal call. Connect every one of them to the assistant first, or plan to watch your CSMs stop using it by spring.
Redesign the work before you add the tool
Weddle, a senior partner at McKinsey and a contributor to the second edition of Rewired, spoke with McKinsey editor Danny McCance for Brooke Weddle on why AI transformation starts with rethinking work, published November 16. She says companies often introduce technology without first reconsidering the work it is meant to support. "You don't want to tech-enable a bad process," she says. In her finance-workflow example, analysts freed from transactional tasks raise new questions about skills, role readiness, and organizational support. She tells leaders that adoption rates show only part of the picture, because usage can decline after launch, and she recommends tracking whether usage keeps creating value. She also flags employee anxiety and the risk that removing transactional work also removes moments of human connection. "Meaning has always been important, and now we need to figure out how that gets constructed in this new environment," she says.
Weddle's argument maps to the people and process legs of the Four-Legged Stool in Coherence at Velocity. Platforms draw most of the budget and most of the attention in a transformation, and people and process decide whether the change lasts past the first quarter. Her point about declining usage gives leaders the test: measure adoption at month five as well as week two.
My take: Pick one workflow per function for 2027 and map it end to end before any tool is selected. Assign a named owner for the redesigned roles as well as the platform, and track usage and outcome together for at least two quarters.
Run the same 2027 plan two ways at a B2B software company
Take a hypothetical B2B software company with $500 million in revenue and a $40 million marketing budget. Agency fees take $9.6 million, or 24 percent. The 2027 plan includes three items: renewing the lead agency, rolling out an AI content and campaign assistant to 150 marketers, and giving 60 customer success managers AI account summaries across 3,000 accounts with $300 million in renewing annual recurring revenue.
In the first version, the CMO renews the agency on its hourly retainer at $9.6 million. The agency adopts AI for production and keeps billed hours flat, because fewer hours would mean less revenue. The marketing assistant reports 41,000 hours saved, and nobody can say where those hours went. The success team's AI summaries reach 92 percent adoption in month one and fall to 40 percent by month five, because the summaries leave out support tickets and product usage, and CSMs go back to their own spreadsheets. Gross revenue retention stays at 88 percent. When the CFO asks what the $1.4 million in AI spend bought, the answer is a number of hours.
In the second version, the CMO and the head of marketing operations rewrite all three definitions of success before signing. The agency moves to a $5.5 million base fee plus up to $2 million in performance components tied to pipeline from agency-built programs and to content cycle time, with a clause requiring the agency to train two in-house teams on its AI production workflow. The agency earns $1.7 million of the performance pool, so total fees come to $7.2 million, or 18 percent of the budget, a saving of $2.4 million. The marketing assistant is judged on experiments run and cost per qualified opportunity: the team goes from 40 tests a year to 120, and cost per qualified opportunity falls from $2,100 to $1,750. The head of customer success operations connects support tickets, product usage, and billing data to the account summaries before launch, and the CSMs redesign the renewal-risk workflow with the tool in it. Usage holds at 85 percent in month five. Gross revenue retention rises from 88 to 91 percent, which keeps $9 million of the $300 million renewing base.
Both versions used the same agency, the same tools, and the same people. The second CMO changed what each contract and scorecard counted.
Take these actions before you sign 2027 renewals
Rewrite agency terms around shared outcomes. Move to a base fee plus performance components, and add a capability-transfer clause.
Make a decision metric the renewal condition for AI tools. Ask each owner for one improved decision and one customer outcome alongside hours saved.
Connect customer context before you deploy assistants. List the sources your best CSMs check and wire every one into the tool.
Map one workflow per function before tool selection. Name an owner for the redesigned roles as well as the platform.
Track usage and outcomes at month five. Treat a launch-week adoption figure as a starting point.
Fund readiness as part of every release
Coherence at Velocity tells leaders to shift measurement from attendance to absorption. Table 10.1 separates release, the evidence that something shipped, from readiness, the evidence that people can use it consistently enough to support the intended customer experience. The manuscript's three-layer frame asks whether the organization is reducing avoidable effort, whether customers encounter less friction, and whether the enterprise is protecting Business Lifetime Value by cutting the recurring cost of workarounds. Wilson's pricing shift, the Forrester team's adaptive advantage, Srebnick's customer progress, and Weddle's month-five usage each measure readiness.
Having sold two companies and acquired three, I can report that diligence teams discount activity reports quickly and spend their time on renewal rates and client tenure. Your CFO will apply the same discount to your 2027 plan.
Answer the opening question line by line. For every agency contract, AI tool, and service program up for renewal, write the customer outcome it is paid to produce, the readiness evidence you will check at month five, and the clause that lets you change terms if neither appears.
Featured this cycle:
Gartner Predicts That By 2030, Agency Fees Will Represent 15% of Marketing Budgets, Down From 24.5% At The Start of the Decade. Jay Wilson, Gartner, November 16, 2026
AI ROI: Move From Minutes Saved To Value Created. Michael O'Grady, Brian Hopkins, Kevin Ogunsua, and Greg Zorella, Forrester, November 18, 2026
AI Is Changing Customer Success But Not Its Mission. Shari Srebnick, Forrester, November 18, 2026
Brooke Weddle on why AI transformation starts with rethinking work. Brooke Weddle with Danny McCance, McKinsey, November 16, 2026