Marketing AI Agents Jump 155% While Data Readiness Slips

For most of the past three years, AI in marketing meant a person with a chat window: someone drafting a subject line, summarising a report or roughing out a brief, then deciding what to keep. That model is giving way to agents, software that takes a task and runs it, often across several systems, with a person checking the output rather than producing it. The latest survey data shows how quickly that handover is happening. It also shows that the inputs agents depend on, clean customer data and a clear way to judge results, are not keeping pace.

Agents Have Moved From Pilot to Payroll

The fourth annual AI and the Future of Marketing report from Basis, published this month and based on a survey of 184 marketing and advertising professionals at agencies and brands, puts agent adoption at 51.1% of organisations. That is a 155.5% increase on the previous year. Everyday AI use has climbed alongside it: 61.4% of respondents now use AI daily, against 33.6% a year ago and 10.9% in 2023, and only 1.6% do not use it at all.

The work agents are being given is revealing. Reporting and analytics is the most common use, at 72.3% of organisations running agents, followed by drafting campaign briefs (56.4%), creating content (45.7%) and planning media (35.1%). Those are not side projects. They are the tasks that decide what a team believes is working and what it does next. Basis also found that 47.8% of respondents replaced human tasks with AI in the past 12 months, up from 27.9% the year before and 10.4% in 2023. Three quarters, 75.0%, say AI made them moderately or significantly more efficient over the past year.

The Data Underneath Is Getting Thinner

An agent that builds a report or plans a media buy is only as good as the data it reads, and on that measure the same survey moves the wrong way. Only 13.0% of Basis respondents describe their organisation's first-party data as extensive and well structured, down from 17.9% a year earlier. In other words, the share of teams handing analytical work to software rose sharply in the same year that the share confident in their own customer data fell. An agent working from fragmented records does not flag the gaps; it produces a confident summary of whatever it can reach.

The budget picture explains part of the mismatch. The 2026 Gartner CMO Spend Survey of 401 marketing leaders, reported by Marketing Dive, found CMOs direct an average of 15.3% of their budgets to AI, while total marketing spend barely moved, at 7.8% of company revenue against 7.7% in 2025. Some 70% of CMOs called becoming an AI leader a key goal for 2026, but only 30% said their organisations have the infrastructure to get there. Those better equipped to scale AI put 21.3% of budget behind it. Meanwhile 56% of CMOs said they lack the budget to execute their strategy and 54% said they lack the resources. Ewan McIntyre, chief of research in Gartner's marketing practice, summed up the gap: "most marketing organizations are not yet built to capture that value."

Review Is Becoming the Real Job

If agents do more of the producing, the human role shifts to checking, and the evidence suggests most teams know it. In the Ahrefs survey of 301 marketers published on September 30, 93% use AI to create or help create content and 85% of those users say they produce it faster. Yet 66% trust AI-generated content only if a human reviews it first, and just 2.7% trust it with no review at all. Some 56% of AI users say human-written content is still better than their AI output.

The direction of travel is still towards less editing. Ahrefs found 7% of AI users now say more than three quarters of what they publish is pure, unedited AI, up from 4%, while the share publishing no pure AI content fell from 50% to 40%. And 78% do not disclose AI use to readers. Measurement is the weakest link of all: Basis found only 6.5% of respondents evaluate AI investments against defined metrics and benchmarks, and 65.1% either do not measure AI return or do not know whether anyone in their organisation does. Half, 50.0%, now call AI's brand safety and misinformation threat significant, a first for the survey.

• Inventory what agents already touch. List every report, brief, content stream and media plan an agent now drafts or runs, and name the person accountable for checking each one.
• Fix the data before widening the remit. With only 13% confident in their first-party data, unify and clean the customer records an agent reads before handing it another analytical task.
• Set the benchmark before the rollout. Record the current time, cost and error rate of a task, then compare the agent against it; without a baseline there is nothing to measure.
• Budget review time as real work. If two thirds of marketers only trust AI output after a human check, that check needs hours on the plan, not leftover capacity.
• Watch the unedited share. Track how much published content goes out without human edits, and set a ceiling for brand-sensitive channels.
• Match spend to readiness. Only 30% of CMOs told Gartner they have the infrastructure to scale AI, so sequence data, governance and skills ahead of new licences.

The agent wave is not waiting for marketing organisations to be ready, and on the Basis numbers it has already reached half of them. The teams that get value from it will be the ones that treat data quality, review and measurement as part of the deployment rather than clean-up afterwards. Everyone else will be automating reports they have no reliable way to check.