AI is good at a lot of things. Generating vendor audit questions when you don't have the context to evaluate the answers is not one of them. Here's what genuine marketing agency accountability actually looks like.
I want to tell you about someone I'll call John.
John runs an independent venue. He'd been working with a solid marketing team for a while, including an excellent developer and a digital ads manager who knew his account well. Things were working reasonably well; the relationship was functional and the results were defensible.
Then John discovered AI. Specifically, he discovered that he could describe his situation to ChatGPT and ask it to help him evaluate his vendors. So he did. He fed it his monthly reports, described his frustrations, explained what he was spending, and asked ChatGPT whether he was getting a good deal.
ChatGPT gave him a lot of questions to ask. Long, detailed, technically framed questions. The kind of questions that, on the surface, looked like serious vendor accountability. John sent them.
The problem was that the questions weren't actually the right questions for his situation. They were generic, generated from a description John had provided. His prompt included his frustrations and suspicions but not the full context of what his vendors were doing or why. Some of the questions misunderstood how the platforms worked. Some accused the vendor of practices that weren't happening. And some were so laden with implication that they couldn't be answered without the vendor feeling they were defending themselves against charges that hadn't been stated directly.
John's developer, who had been doing excellent work for years, got a message so hostile in tone that he stepped away. The ads manager, who had real knowledge of the account, disengaged rather than try to respond to questions that didn't map to reality. John was left with a clean slate and no team, which he experienced as a victory right up until he had to rebuild from scratch.
I'm sharing this story because I've seen versions of it more than once. AI is a powerful tool for a lot of things. Generating vendor audit questions when you don't yet have the foundation to evaluate the answers is not one of them.
"The questions looked like serious accountability. They weren't the right questions for his situation, and he couldn't evaluate the answers he got."
WHAT AI ACTUALLY CAN'T DO IN THIS CONTEXT
AI can tell you what questions a sophisticated client might ask a marketing agency in general. It cannot tell you whether your specific vendor is doing those things well, poorly, or at all, because it doesn't have access to your account, your industry benchmarks, your conversion data, or the context behind the numbers in your report. It's working from what you tell it, and what you tell it is shaped by what you already believe.
If you're frustrated with your vendor and you feed that frustration into an AI prompt, the output will reflect your frustration back at you in formal language. That's not analysis. It's amplification.
THE QUESTIONS THAT ARE ACTUALLY WORTH ASKING
Genuine marketing agency accountability doesn't require AI. It requires knowing what your campaign is supposed to accomplish and whether the report in front of you reflects progress toward that. A few questions that belong in every monthly review, regardless of channel:
- What is the campaign optimizing toward, and is that the action I actually care about? (Clicks and impressions are not the same as ticket sales or inquiry submissions.)
- What did we spend last month, what did we get for it, and how does that compare to the prior period and to the same period last year?
- What changed in the campaign this month, and why?
- What's not working, and what's the plan for it?
- Are there things I could look at directly if I wanted to verify this myself?
That last question matters. A good vendor will say yes and show you where. A vendor who can't answer it, or who responds with defensiveness rather than transparency, is telling you something worth knowing.
WHEN SOMETHING GENUINELY IS WRONG
Sometimes the concern is legitimate. Vendor underperformance is real; so is billing for work that wasn't done, reporting that obscures rather than reveals, and campaigns built around the vendor's ease rather than your results. Those things happen and they deserve to be addressed directly.
The right response to a genuine concern is a direct conversation: "Here's what I'm seeing, here's what I expected, help me understand the gap." Not a list of AI-generated audit questions. Not a formal review document built on assumptions. A conversation, with your actual numbers, with someone who has access to the same account you're paying for.
If you genuinely can't tell whether your marketing agency is doing good work, that's a real problem, but the solution isn't to weaponize AI against them. The solution is to get a second opinion from someone who can actually look at your account. A one-time audit from a qualified independent reviewer will tell you more than a hundred AI-generated questions.
Post 12 steps back from the vendor relationship and looks at the bigger picture: why marketing is the number one challenge for independent venues nationwide, and what's actually starting to change.