AI Gets a Meter, and Marketers Will Cut a Third of Their AI Initiatives

For most of the past three years, AI in marketing was bought like any other software: a licence, a seat count and a renewal date. That is changing. Vendors are moving to pricing that charges for what the software actually does, so every prompt, every generated asset and every step an agent takes shows up on an invoice. The predictions season has now put a number on what happens next, and it is not more AI everywhere. It is a cull, decided by whoever reads the bill.

The Blank Cheque Is Being Withdrawn

In its 2027 B2C marketing, CX and digital predictions, published on October 21, Forrester forecasts that rising AI costs will push B2C marketing leaders to scale back one third of their AI initiatives. The mechanism is pricing. Usage-based models, the firm says, turn each agent interaction, generated asset and automated workflow into a billable event, so a pilot that looked cheap at small volume becomes a material line item once it runs at the scale it was built for.

The firm's CMO predictions add the survey evidence. In Forrester's Q3 2026 CMO Pulse Survey, 82% of US B2C marketing executives expect to provide stronger justification for future AI investments, and 78% say executive leadership now demands stronger evidence of value than it did 12 months ago. Forrester expects marketers to concentrate budget on use cases with measurable business value, and procurement and finance to take a bigger role in AI decisions. Its summary is blunt: AI will get funded in 2027 because it performs, not because it is new. "The challenge for leaders in 2027 will be gaining greater control without limiting growth," said Sharyn Leaver, Forrester's chief research officer.

The Meter Is Already Running

The shift Forrester describes is not hypothetical. The Gartner 2026 CMO Spend Survey of 401 marketing leaders, reported by Chief Marketer, found 56% of respondents increased the share of their martech budget allotted to consumption-based pricing over the past year, against just 9% who reduced it. Half of the organisations already running consumption-based tools say they continually renegotiate contracts to avoid unexpected usage and cost spikes. Some 41% have set up, or are setting up, real-time usage controls, and 24% are overhauling systems specifically to reduce usage.

That is happening inside a shrinking envelope. The mean share of marketing budget going to martech has fallen to 19.4%, a five-year low, from 26.6% in 2021, even though 62% of CMOs still plan to invest more in it. AI already takes an average of 15.3% of marketing budgets, and 21.3% among the organisations Gartner rates as mature. A fixed or falling martech budget, a growing AI share and a variable price per action is the arithmetic behind Forrester's forecast.

Overruns Are Common, and Pauses Are Rare

The wider enterprise picture shows how organisations behave when the meter runs ahead of plan. Futurum's 2H 2026 survey of 1,636 technology decision makers found 46.9% report AI spend above plan, 35.6% moderately and 11.3% substantially. Only 31.8% are roughly on budget, 5.6% are below it, and 10.0% have no formal AI budget to measure against at all.

What those over-budget organisations do next matters for marketers. Among them, 47.6% ask for more budget and 43.3% absorb the overrun and settle it later, while only 17.2% pause or reduce the AI initiative. Some 23.1% reallocate money from a non-IT business unit budget, which is often where marketing sits. "The budget has stopped being a control," said Mitch Ashley, who leads Futurum's CIO practice. For a CMO, that means AI cost can be cut from marketing's plan by someone else's overrun, and the use cases with the weakest evidence will be the first to go.

Forrester's other predictions point the same way. It expects AI agents to replace legacy marketing and insights tools at one in 10 companies, and it found 64% of US B2C marketing executives believe their AI use is likely to cause a brand-damaging incident within the next 12 months. It also predicts that half of brands will make human-generated content a top priority again as AI output floods digital channels. A cull, in other words, is not a retreat from AI. It is a sorting of the initiatives that can prove their worth from the ones that cannot.

• Price every use case per action. Convert each AI workflow into a cost per prompt, asset or agent task at full production volume, not pilot volume, before the renewal arrives.
• Attach a business outcome to each line. With 78% of leadership teams demanding stronger evidence, any initiative without a revenue, cost or speed measure is the likeliest to land in the third that gets cut.
• Put usage controls in the contract. Half of consumption-based buyers keep renegotiating; caps, alerts and committed-use tiers are cheaper to agree at signature than after a spike.
• Bring finance in early. Forrester expects procurement and finance to gain influence over AI decisions, so build the business case with them rather than defending it to them.
• Retire what agents replace. Where an agent now does a legacy tool's job, decommission the tool, so AI spend funds itself instead of stacking on top of an existing licence.
• Keep humans where the brand is exposed. With 64% of executives expecting an AI-driven brand incident, protect review time for customer-facing content rather than cutting it to pay for more usage.

The first wave of marketing AI was judged on what it could do. The second will be judged on what each use of it costs and returns, and that judgement will increasingly be made with finance in the room. Teams that measure value per action now will choose which third of their AI portfolio to cut. Everyone else will have that choice made for them.