Our Take
The real story isn't that AI agents can do marketing work—it's that the vendors publishing case studies haven't yet shown what fails when humans step back.
Why it matters
Marketing teams are moving beyond AI-as-tool to AI-as-collaborator, but most evidence comes from vendor-selected examples. Practitioners need to know failure modes, not just wins.
Do this week
Marketing leaders: audit your current AI vendor case studies for explicit mention of failure rates, human override frequency, and cost per campaign before committing headcount.
Acquia and Gartner Frame Agentic Marketing as Human-AI Partnership
Acquia, a content and experience platform, has published a Gartner-backed report positioning AI agents as collaborative tools for marketing teams rather than full replacements. The framing centers on "real change" driven by humans and machines working together, with examples of how organizations are using agentic AI to handle specific marketing workflows.
The report does not reveal its source material (independent case studies, customer interviews, or vendor examples), nor does it publish benchmark data on success rates, cost per outcome, or the frequency of human intervention required to correct agent decisions.
Vendor Case Studies Without Independent Baseline
Marketing teams considering agentic AI deployment face a familiar pattern: vendor-published examples showing collaboration scenarios, but no independent data on when that collaboration breaks down. The Gartner partnership lends credibility to the framing, but Gartner's own 2024 AI adoption research has consistently noted that most enterprise AI deployments require significantly more human oversight than their initial proposals estimated.
Without stated metrics on agent accuracy, false-positive rates in customer targeting, or the cost of human review cycles, the "real change" claim remains anecdotal. Marketing budgets are material. Decisions made on this framing could shift headcount allocation from analysts to oversight roles without net productivity gain.
What to Demand Before Adoption
Request specific data before piloting agentic marketing tools: (1) error rate per task type (e.g., audience segmentation, email copy generation, campaign timing); (2) percentage of agent outputs that required human revision in the case studies cited; (3) total cost per campaign including both agent compute and human review labor; (4) comparison to your current manual workflow cost baseline, not to the vendor's estimated time savings.
If your vendor or analyst report cannot provide those numbers, you are evaluating marketing theater, not marketing capability. Ask for anonymized customer audit data, not curated success stories.