Questions to Ask Before Hiring Any Marketing Agency in the AI Era
By Stacey Lievens · 2026-07-10 · 5 min read
Hiring a marketing agency used to mostly be a question of taste and price. In an environment where AI-generated content is cheap and audiences are increasingly skeptical of it, a few additional questions matter as much as either.
1. "How much of the content you produce is AI-generated, and how is that disclosed?"
This isn't a trick question, and a good agency shouldn't treat it as one. AI-assisted editing and organization is fine. The concern is content presented as coming from a real customer or expert voice that was actually generated wholesale. Given that over half of consumers reduce engagement the moment they suspect AI involvement, an agency that can't clearly answer this question is a risk to your Trust Gap™, not just your budget.
2. "Can you show me a case study with a named client and a specific, checkable outcome?"
Vague results ("significant improvement," "great engagement") are a warning sign. A specific number tied to a named client, with a describable mechanism for how the result happened, is the standard your own marketing should be held to, and it's a reasonable standard to hold a prospective agency to as well.
Pay attention not just to whether they can produce one, but how quickly. An agency that has to dig for a real example, or that offers a story with the client's name redacted "for privacy," is telling you something about how central real proof actually is to their own way of working, regardless of what their pitch deck claims.
3. "What's your process for collecting real customer testimonials, not writing them?"
Some agencies still produce testimonial-shaped copy for a client to "approve" rather than collecting an actual customer's actual words. That's Manufactured Messaging™ regardless of how polished it reads, and it carries real risk if it's ever discovered to have been invented rather than collected.
Ask specifically who conducts the interview, whether the customer sees questions in advance, and whether raw footage or a transcript is available for your own review. A clear, confident answer to all three is a good sign. Hesitation on any of them is worth probing further before signing anything.
4. "How do you measure success beyond vanity metrics?"
Impressions and engagement are easy to inflate and don't necessarily reflect trust. Ask specifically how the agency tracks whether content is actually reducing the Trust Gap™, shorter sales cycles, more specific objections answered, prospects citing a particular story in a sales call, not just how many people saw a post.
An agency that reports only reach and engagement, without ever connecting the work to sales cycle length, conversion, or specific deal outcomes, is optimizing for numbers that look good in a monthly report rather than numbers that reflect whether trust is actually being built.
5. "What happens to my customer stories and data if we stop working together?"
A library of real customer stories is one of the most valuable assets a business can build. Make sure ownership and portability are clear before signing, not after.
This question is easy to skip during an initial sales conversation focused on excitement about the work ahead, which is exactly why it's worth asking early and in writing. A business that has spent a year building a deep, specific proof library doesn't want to discover mid-transition to a new agency that the raw footage, transcripts, or usage rights aren't fully theirs.
What a Strong Answer Actually Sounds Like
It helps to know what a good answer sounds like, not just what a bad one looks like. On AI disclosure, a strong agency will say something specific: "We use AI to help draft first passes and organize research, but every testimonial and case study comes from a real interview we conduct, and we'll show you the raw footage or transcript." On case studies, a strong agency can walk you through the actual mechanism behind a result, not just the headline number, because they were close enough to the work to understand why it happened. Vague, deflecting answers to specific questions are themselves useful information, regardless of how confident the delivery sounds.
What to Do With the Answers You Get
Treat this less like a pass/fail test and more like a diagnostic. An agency that stumbles on one question but answers the other four with real specificity and evidence might simply need a clarifying follow-up conversation. An agency that deflects on most of them, or answers every question with confidence but no actual example to point to, is telling you something important about how the relationship will likely go once the retainer starts and the pressure to produce visible activity, rather than real proof, sets in. Write down the answers. Compare them across the two or three agencies you're evaluating. The differences tend to be more revealing than any single pitch deck.
Red Flags Beyond the Five Questions
A few additional signals are worth watching for during the sales process itself, before a single deliverable arrives. An agency that can't name a single client willing to serve as a reference is worth a second look, given how central real proof is to their own pitch. An agency whose own marketing consists entirely of confident claims with no attributed client stories is, in effect, failing to apply its own advice to itself. And an agency that treats these five questions as adversarial rather than reasonable due diligence is signaling something about how they'll respond to scrutiny later in the relationship.
Why These Questions Matter More Than They Used To
In a market where AI has made polished content nearly free, the differentiator isn't how good an agency's copy sounds. It's whether they're building something real: a library of specific, attributed proof that compounds in value over time, rather than a stream of content that reads well but leaves nothing durable behind once the retainer ends.
Frequently Asked Questions
Why does it matter how much of an agency's content is AI-generated?
Because audiences increasingly reduce engagement with content they suspect is AI-generated, so an agency's approach to disclosure and where AI is and isn't used directly affects whether their work helps or hurts your Trust Gap™.
What's a red flag when an agency shows you case studies?
Vague, unattributed results like 'significant improvement' without a named client or specific, checkable number. Strong case studies name the client and the mechanism behind the result.
Should an agency ever write testimonials on a client's behalf?
No. A testimonial written by an agency rather than collected from a real customer is Manufactured Messaging™ regardless of how polished it sounds, and it carries real risk if discovered.
Who should own the customer stories an agency helps collect?
The business itself. Ownership and portability of any collected customer stories should be clarified in the contract before signing, since this library is a durable asset regardless of the agency relationship's outcome.