Ask Your Marketing Agency About AI: Four Questions Before You Renew

The renewal call where nobody asked what AI meant

Say you run marketing at a 60-person software company and your agency’s renewal is three weeks out. The account director’s deck has a new slide titled “AI-enabled delivery,” with three bullets: faster turnaround, more variants, smarter insights. Everyone nods. Nobody asks what “AI-enabled” means, because asking feels like accusing them of something.

Three months later the blog drafts are arriving in two days instead of six. One of them says your product integrates with a tool it doesn’t integrate with, and your product marketing manager catches it by luck on a Friday afternoon. The retainer is exactly what it was: 48 hours a month of “content and strategy.”

None of that means the agency did anything wrong. It means the AI question didn’t get asked on purpose. Four questions cover most of it: what does AI touch in the work you buy, what are the hours buying now, who checks the output, and who owns it and sees your data. The rest of this guide is how to ask each one, and what a good answer sounds like.

The questions work whether you’re a head of marketing deciding on a renewal, a brand manager who inherited the agency relationship, or a performance marketer who wants to know whose numbers are in the monthly summary. They also work for freelancers, in a shorter form (the FAQ covers that). An agency using AI is often good news for you. What I’m arguing for is knowing about it, in writing, before something goes wrong.

Ask what AI touches, deliverable by deliverable

“AI-enabled” can mean a transcription tool on the weekly call or a language model writing your ad copy, and those deserve very different conversations. So skip “do you use AI?” The answer is usually yes, and it tells you nothing. Ask for a one-page map instead: for each thing you pay for, where AI is involved and which human looks at it afterwards. Here’s a good answer, filled in for the same agency.

I’d read the last column first. A prompt containing your pricing, an unreleased feature or a customer list is a different risk from one containing your published brand guidelines, and you only learn which you have by asking. Ask for the map before the renewal meeting, not in it. An account lead who knows the answers can write it in about an hour. If it takes noticeably longer, that’s useful information too.

Blank cells are fine at first. They’re your agenda for the renewal call.

The map also does a quieter job. In June 2023 the Association of National Advertisers revised its media agency contract template to require an advertiser’s prior consent before an agency uses AI applications in delivering its services. As trade coverage noted at the time, the template doesn’t define what AI is.^[1] That gap is worth closing in your own paperwork, and a deliverable-by-deliverable map gives you a working definition in the meantime.

Ask what the hours are buying now

Many retainers are priced on labour, and AI changes how much labour a deliverable takes. The Association of National Advertisers’ own summary of its 2026 agency compensation work says most marketers are satisfied with how they pay agencies, yet many plan changes amid concerns about transparency, rebates and AI-driven labour.^[2] The full findings sit behind a member login, so I can only point to that summary, not the numbers beneath it. Here’s the retainer from the opening, before and after, with the hours filled in.

There are three straightforward answers to where the hours went, and all three can be fair.

• Lower price. The saving comes back to you as a smaller fee.
• More output. Same fee, more variants, tests or formats, which helps only if you wanted more of them.
• More judgment. The hours move into strategy, testing and review that you can see in the work.

Silence is the answer to worry about.

An agency that has built real workflows and real review steps has a legitimate case for keeping its fee, because you’re buying judgment and outcomes, not typing speed. It should be able to say what it’s buying you instead of hours. If you’d rather price by deliverable than by the hour, that’s a normal conversation. And if the numbers have you wondering whether a function belongs in-house, our piece on replacing a $16K agency quote shows what that looks like at the smallest scale.

Notice that this is the one question where I wouldn’t open with a discount. Open with “what changed in how you work, and what should change in what we’re buying?” The first version tends to get a defensive answer and the second tends to get a plan. That’s our judgment about how these conversations go, not a measured result.

Ask who checks it, and what they check

Say you’re the brand manager and the agency’s launch email lands in your inbox with two things you wouldn’t approve: a line calling you “the industry’s most trusted platform” and a customer quote you don’t recognise. You ask who wrote it. The answer, after a pause: a draft from an AI tool, tidied by a junior copywriter on deadline, and the quote was meant to be a placeholder.

The placeholder matters more than it sounds. In August 2024 the US Federal Trade Commission announced a final rule banning fake reviews and testimonials, including AI-generated ones that misrepresent being by someone who doesn’t exist.^[3] A made-up quote in a customer email is your brand’s problem, not only the agency’s. The fix is a named person for each deliverable and a check you can see, rather than banning the tool. Here’s the split.

A name, not a team.

“Our team reviews everything” is not a name. It’s the sentence you hear from a careful agency and a careless one alike, which is exactly why it can’t be your evidence. Ask for a person and for what that person looks at: the facts, the numbers, the voice, the legal claims. In a regulated category such as financial services or health, add your own legal or compliance reviewer to the right-hand column, because the agency’s reviewer isn’t accountable for your sector’s rules.

Then run the spot check. Pick one deliverable a month at random and ask for the source behind every claim in it. Random matters: a check you only run when you feel suspicious teaches the agency when to be careful. For the disclosure and pre-publish side, including the FTC rules that already apply, our AI marketing compliance checklist covers what to check before anything goes live.

Ask who owns it, and where your data goes

Two different questions hide in here. The first is ownership, and it’s less settled than most contracts assume. In January 2025 the US Copyright Office concluded that, with current generally available technology, prompts alone don’t give a person enough control to be the author of what comes back. Human authors keep copyright in what they contribute, such as their own expression or the creative selection, arrangement or modification of AI output.^[4]

For a marketer that has a practical edge. A logo, mascot or hero illustration that is essentially a generated image might not be protectable the way you’d assume, so ask what a human designer actually did to it. Whether that matters for your brand is a question for your own lawyer. I’m describing what the Office said, not giving legal advice.

Then there’s the data.

The Association of National Advertisers and the law firm Venable published an AI Rider template in July 2025 to help members draft and negotiate AI use in creative services.^[5] Trade coverage of it describes provisions on telling the client when AI is used, some human oversight of AI-generated work, ownership of commissioned output, and transparency about how tools were trained and which third-party tools are involved.^[6] The template sits behind member access, so I’ve worked from that coverage, and your counsel would need to turn any of it into contract wording.

You don’t need the rider to start. You need answers to a few plain questions, and an email is a fine place to ask them. Here’s one to send your account lead this week, with the brackets filled in.

Score the answers

Answers are easier to compare when they’re scored. This is a five-line sheet we made for the purpose, not a standard, and the bands are our judgment. Score each row 0, 1 or 2 after the call, using the agency’s written answers where you have them.

Score it after the call, not during it.

Different roles will care about different rows. A head of marketing or CMO usually owns rows two and five, because they decide the fee and the contract. A brand manager or content lead is closest to rows one and three. Procurement or finance will push on row four. A performance marketer should read the monthly summary line in the map very closely, because numbers that nobody recomputed are the ones that end up in a board deck.

Check it's still true in three months

A good answer on the call is the start. What you want to know three months later has three parts, and they’re the same Use, Persistence and Impact lens we apply to any AI rollout.

• Use. Are the AI-assisted deliverables flagged in each month’s handover, as the map says they would be?
• Persistence. In month three, with nobody reminding them, is the named checker still checking?
• Impact. Is the work faster, better or cheaper for you? Look at revisions per deliverable, turnaround and results, not the number of AI variants produced.

Counting variants is a usage number. It tells you the tool is switched on, not whether you bought anything. Here’s a register that takes ten minutes a quarter, with two rows filled in.

One more thing sits on your side of the table. Whoever joins these calls needs enough hands-on AI use to judge the answers, because “we prompt it and then review it” sounds identical from a careful agency and a careless one. Training builds that capability. Workflow design gives it somewhere to land, which here means a review step in your own monthly rhythm. Reinforcement is the quarterly check turning that step into a habit. A team can start with the answer sheet on its own, and a structured workshop shortens the learning curve and puts the whole team on the same standard, rather than leaving it with one enthusiastic manager. That’s the difference we describe in what makes an AI-powered professional.

So start small. Send the email above to your account lead this week, put the answers next to the contract, and score them at the next renewal conversation.

Hina Mian, Co-Founder of Future Factors AI

Hina is a marketing strategist with over a decade of hands-on campaign experience across B2B and consumer brands. She writes about using AI to run leaner, sharper marketing without losing the human touch. Future Factors helps professionals and teams build practical AI capability through role-based training, workflow design, and hands-on adoption.

More about Hina →