Finding 1.1
Vendor claims versus reality
A perceived need for human oversight means agents only partially live up to marketing claims
say agents only partly live up to the claims: they’re useful for bounded tasks but need heavy oversight
Marketers want faster execution, but not at the cost of control over what customers see. Three in five (61%) say agents only partly live up to the claims made for them: they help with bounded tasks, but teams still keep a close eye on them instead of handing off whole workflows. Only 7% say agents fall short outright. Company size doesn’t change the picture much: 65% of mid-market and 62% of enterprise marketers say agents only partly deliver.
Marketers see real value in speeding up well-defined work, and they want a person reviewing anything that shapes a marketing decision. For vendors and marketing leaders, that’s a clear signal: transparent, controllable help beats promises of hands-off, end-to-end autonomy. Regulators have drawn the same line: the FTC’s Operation AI Comply targets companies that overstate what AI can do.
-
01
30%
Human judgment remains the boundary
The claim marketers doubt most (30%) is that agents can replace human marketers, judgment, and emotional intuition. Their concerns go beyond the quality of the output.
-
02
34%
Brand handoff depends on explicit inputs
About a third (34%) say agent drafts capture brand tone well when teams give them detailed guidelines and examples, which makes good inputs a practical condition for wider use.
-
03
52%
Mid-market teams do more of the editing
More than half (52%) of mid-market marketers say agent drafts need human editing to sound on-brand, compared with 36% at enterprises.
For the most part, [agents] live up to the claims, but completely eliminating a human review from an expert would not be wise and not something that I would fully trust at this stage. There are certain lower level functions that are really just sharing information that I would be more likely to trust AI with. then generating language or content that needs to be strictly regulated.
Depends on what kind of claims we're talking about. If you hear the ones the hyped ones that agents can do everything from A to Z, I don't think we see that. I think products that have to interact that have to do that bridge between the provider, the vendor and the customer still need a human in the loop.
Finding 1.2
AI Agents Living Up to Claims by Marketing Role
Across lifecycle, operations, brand, and product and growth roles, 60% to 82% see agents as useful but oversight-heavy. The mixed Other group is the exception. Build review gates that fit each function instead of assuming autonomy carries over from one team to the next.
View data table
| Category | Yes, lives up to claims for speed and productivity | Partially; assists with bounded tasks but requires heavy oversight | No, falls short of autonomous end-to-end marketing claims | Other |
|---|---|---|---|---|
| Lifecycle, CRM, and email marketing (n=15) | 4 | 9 | 2 | 0 |
| Marketing leadership, generalist, or multi-channel operations (n=44) | 11 | 28 | 3 | 2 |
| Brand, content, and communications strategy (n=16) | 4 | 10 | 0 | 2 |
| Product, growth, and acquisition marketing (n=11) | 2 | 9 | 0 | 0 |
| Other (n=14) | 2 | 5 | 2 | 5 |
