BNT AI Partners Blog
Forrester data shows a double-digit satisfaction gap across every AI capability marketing leaders say they want from agency partners. Here is what is actually failing, and the five questions that separate good agencies from expensive mistakes.
Forrester's research on agency AI capabilities is worth sitting with for a moment. Across every AI-related capability that marketing leaders say they want from agency partners, there is a double-digit gap between how important the capability is rated and how satisfied clients actually are with what they receive. Not in one or two areas. In every area that matters: strategy alignment, brand voice integration, content quality, and measurement.
This is not a complaint from a handful of demanding clients. It is a pattern that has been building as agencies rushed to position themselves as AI-first without building the competency to back it up. If you have worked with an agency that uses AI and felt like something was off but could not quite name what it was, there is a good chance you experienced one of three specific failures. Here is what they are and how to tell whether the next agency you consider has figured them out.
The Forrester data puts a number on something a lot of marketing owners have felt but struggled to articulate. When they are asked what they want from an AI marketing partner, they consistently name things like strategic integration, brand voice fidelity, and measurable business outcomes. When they are asked how satisfied they are with those same capabilities from their current agency, satisfaction falls 10 to 20 percentage points behind importance on each one.
That gap is not catastrophic failure. It is something harder to see: work that looks acceptable on the surface while consistently missing what you actually care about. An agency that produces content on schedule, responds to feedback promptly, and keeps the retainer running smoothly can still be failing you on every dimension that moves your business forward. The gap is expensive precisely because it is easy to miss until you do the math on what you spent versus what changed.
The agencies closing that gap are not necessarily larger or more expensive. They are doing three specific things differently from the ones producing the gap.
The first failure is using AI for speed instead of strategy. Most agencies that adopt AI tools use them to produce content faster. The brief is still written the same way, the strategy is still the same, and the output is generated more quickly than before. Speed is a real benefit, but when it becomes the primary value proposition, you get more content that is mediocre rather than less content that actually works. Faster production of the wrong thing is not progress, and it is the most common thing being sold right now.
The second failure is not integrating AI with the client's actual voice or data. Generic AI output sounds generic because it was built on generic inputs. Agencies doing this right build client-specific context into their systems: documented brand voice, real customer language pulled from reviews and sales calls and intake forms, and ICP profiles grounded in actual buyer data. Agencies that skip this step produce content that could have been written for any business in any industry. You can usually tell within one read. If a sample could plausibly belong to a competitor, the agency has not solved the problem that matters.
The third failure is measuring outputs instead of outcomes. An agency reporting on number of posts published, word count, or email open rates is measuring what they produced, not what changed because of it. Agencies focused on outcomes can point to leads generated, pipeline influenced, or conversion rates that moved. If an agency cannot draw a clear line from their work to those numbers, they are optimizing for their own production efficiency rather than your results. This one is the easiest to spot in a proposal: look at what they promise to report on.
"Faster production of the wrong thing is not progress. If speed is an agency's primary AI value proposition, you will get more content that does not work, delivered faster."
The agencies worth working with build a client-specific content system before they produce anything. This means documented brand voice, ICP profiles built from real customer data, and a content strategy that ties each piece of output to a business goal. AI runs inside that system, not instead of it. The system is what makes the output consistent, on-brand, and actually useful. Without it, you are just getting fast drafts.
They show you how they measure results before you have to ask. Not impressions or engagement rates as the primary story. A clear line from the work they produce to the numbers that matter to your business — leads, pipeline, conversion rates, revenue attributed. If that conversation is uncomfortable for them before you sign, it will be uncomfortable every month after too.
They produce marketing you would not immediately identify as AI-generated. This sounds like a low bar, but most AI-assisted marketing is still instantly recognizable as AI-assisted. Agencies that have genuinely figured this out use AI in the workflow without letting it flatten everything into the same rhythm and structure. If you read a sample and know within two paragraphs that a language model wrote it, the agency has not solved the problem that their buyers are noticing. You can read more about what makes AI content credible to buyers in the post on why your AI content might be hurting you.
Ask them to show you how they document client brand voice before they start writing. If the answer is a short intake form with a few questions, they are working from surface-level inputs. A real brand voice system includes documented tone guidelines, vocabulary choices, things the brand never says, and ideally, examples from actual customer language. If they cannot show you that system, they do not have one.
Ask what data they use to inform their AI systems for each client. If the answer is "our prompts" or "our experience with your industry," they are working from generic inputs. The right answer involves actual client materials: sales call transcripts, customer reviews, existing high-performing content, and real buyer language. That data is what separates output that sounds like you from output that sounds like everyone else.
Ask to see a content draft that was revised or rejected alongside one that was used. If every draft comes out clean on the first pass, they are either not sharing the real picture or they are publishing without meaningful review. Good AI-assisted work involves revision. If there is no evidence of that process, there is probably no real quality standard in place.
Ask how they measure the business impact of their work, separate from production metrics. Get specific: ask them to walk you through a client whose results they are proud of and explain which numbers changed and why they credit the content work. If the best they can give you is "we helped increase their social presence," that is a production metric wrapped in outcome language.
Ask whether they have ever pushed back on a client's content direction because it would not perform. Agencies genuinely focused on your outcomes have opinions about strategy. They will tell you when a campaign idea will not move the right buyers, when a topic is too generic to get traction, or when the messaging is not aligned with what your customers actually care about. Agencies focused on output just execute what they are handed. That difference shows up in the answer to this question. For a framework on how to measure what good AI marketing actually produces, the post on how to calculate the ROI of AI covers that in detail.
Clear, specific answers to these five questions mean you have found an agency worth continuing the conversation with. Vague answers, repositioning, or discomfort mean you have saved yourself the cost of finding out the hard way.
A good AI marketing agency builds a client-specific content system first, including documented brand voice, real customer data, and a strategy tied to business goals. AI runs inside that system to produce content at scale, but the strategy, voice, and measurement are human-driven. The output should be indistinguishable from well-written human content and traceable to business results.
Ask them how they measure success. If they lead with production volume, turnaround time, or cost per piece, speed is their primary value. If they lead with pipeline influenced, conversion rates, or lead quality, they are building toward results. What they prioritize in the first conversation is almost always a reliable signal of what they will optimize for once you sign.
Ask how they document client brand voice, what data they use to inform AI outputs, how they review drafts before delivery, how they measure business impact rather than production volume, and whether they have ever pushed back on client direction for strategic reasons. Specific answers to all five separate agencies that have figured it out from those that have not.
Outputs are what was produced: posts published, emails sent, words written. Outcomes are what changed because of that work: leads generated, sales conversations started, conversion rates improved. An agency measuring outcomes can draw a direct line between their work and your business results. An agency measuring outputs is tracking their own productivity, not yours.
Ask for writing samples from current clients in your industry and read them out loud. If they sound like a language model wrote them, they will read that way to your buyers too. Ask to see a draft that was revised or rejected alongside one that was used. If every draft comes back clean on the first pass with no revision history, there is likely no real quality review happening before content goes out.
If you have worked with an AI marketing agency and felt like something was off, or if you are trying to figure out how to evaluate your options before committing budget, we are glad to take a look at what you are working with. The assessment is free, and we will give you an honest read on where things stand.
Get your free assessment