Man in a red plaid shirt and red crocs sits in a glowing tree on a dark suburban street at night as five rusted robots stand in a row below, symbolizing a real creator as the new creative agency.
October 12, 202610 min read

Your Next Creative Agency Is a Creator

You probably saw it the way I did. Late on a weeknight, thumb moving, a Forbes CMO Summit clip slides in between a dog in a raincoat and a guy pressure washing his driveway. A man holding a microphone says brands are going to use creators as their creative agencies.

If you run marketing for a brand, that clip lands one of two ways. Either "finally," or "great, now I am supposed to swap a firm with a boardroom for a guy with a ring light."

Both are fair. The speaker was Arthur Leopold, who founded Agentio, a platform that matches brands with creators selling ad space. So he has a horse in this race. So do I. I make creator content for brands for a living. Weigh us both accordingly, then look at where the money is going, because the money does not care what either of us thinks.

The budget moved before the keynote did

The IAB, the trade association for digital advertising, has US creator ad spend on track for $37 billion in 2025, growing about four times faster than media overall. Close to half the buyers it surveyed now treat creators as a must buy, ranked behind only social and paid search, and the same buyers say they want better attribution tied to business results.

At Cannes this summer, Bose's CMO told Digiday he has not used a creative agency in four or five years. The brand's own content arm, Bose Studios, already makes up a third to 40% of its marketing. And the Association of National Advertisers reports 82% of its members now run an in house agency, up from 42% in 2008.

So the traditional agency is getting squeezed from both ends. Big brands pulled strategy and media inside. Creators are taking the making. The brands stuck in the middle are the mid sized ones, with no in house team and no agency retainer, who still need every job an agency used to do. They just stopped having anyone assigned to it.

What the agency did that a video does not

When a brand hires an agency, it is buying a list of jobs. Somebody figures out who the buyer is and what they are missing. Somebody writes the brief. Then the creative gets made, tested, measured against a goal, kept legal, and somebody answers the phone when it goes sideways.

When a brand hires a creator, it usually buys the video. One file, one caption, one round of notes. The rest of the list quietly evaporates, and six months later the meeting ends with "we tried creators, it didn't really work for us."

I came up in software, not film school. Twenty five years in it, a company built and sold before I turned thirty, and a lot of those years managing people. Management teaches you one thing fast: the work nobody sees decides whether the work everybody sees pays off. Same job here. Better lighting.

Hand over the problem, not the script

If you are going to treat a creator like an agency, treat them like one. You would never hire a creative shop and then dictate the jokes.

A study in the Journal of Marketing, one of the first to fold in what brands paid, found that an influencer's originality and how clearly the brand shows up in the post both raised effectiveness, while posts announcing a new product pulled it down.

In plain terms: the brand needs to be unmistakable, and the creator needs to be unmistakably themselves. A press release read out loud by a guy in a hoodie is the worst of both.

That is why my pitches open on a scene and not a spec sheet. A big tree in suburbia, the camera goes up, and I am sitting in it with your product because the kids took over the house and I needed one quiet minute. People stop. They rewind. The product is the thing they remember. The brand owns the frame and the idea is mine, which is the balance I wrote about in what makes a creative ad work. It is also the part an AI avatar cannot fake for you.

A rusted boxy robot and a human hand in a red plaid sleeve both reach for a glowing vintage camera on a wet backyard bar at night, a beer and a pint beside it, symbolizing who owns the creative.
Everyone wants the camera. The idea is the part you can't reach for.

Name the goal before anybody hits record

Views are the number everyone screenshots and the one that tells you the least. I run about four million a month across five platforms, and it sits near the bottom of my dashboard for a reason. The scoreboard is whatever you are trying to buy: sales, orders, bookings, sign ups. Brand lift (recall, consideration, purchase intent) is the leading indicator that feeds the sale, and you want it moving, but the sale is where you keep score.

A real example from my own data. For Pita Mania, a pita spot in Tecumseh, I ran three versions of the same seven second reel through Instagram trials. Version C pulled 799 views and the best skip rate of the three, 21.9%. Version A pulled 295. On views, C wins in a walk.

The audience column tells a different story. Only 10.5% of C's audience was Canadian. Version A was 74% Canadian. Roughly 84 Canadians saw the "winner." About 218 saw the "loser." A view from Texas does not buy a pita. If the goal was proving the creative works, C won. If the goal was people in Windsor Essex ordering lunch, A quietly beat it.

An agency asks which goal before the test. Most creator deliverables never ask at all, which is how a brand ends up celebrating a reel that fed half of Texas and sold nobody lunch.

One round is a coin flip with a logo on it

Testing is where a creator earns the agency label or loses it.

I learned this in usability labs long before I owned a ring light. You watch a stranger fail to find the checkout button, you fix it, and you bring in the next stranger. Nobody ships after session one.

For Pita Mania that meant seven variations over a week, each one adjusted from what the previous one showed. The winner opens zoomed out on me hauling a safe, then reveals the restaurant, and it landed a 27.2% skip rate against 37.9% for the safe open cut. The catch went in the client deck in plain words: trials go wherever Instagram sends them, and only about 16.6% of viewers across the test were Canadian. The result is directional. It earns a second round with paid targeting around Windsor Essex. Nobody popped champagne.

Trials keep you humble in other ways. A close up of my own teeth went out to a cold audience that was 99.9% non followers, about a quarter of them in India, half of them 18 to 24. For one afternoon I was very big with 22 year olds in Mumbai, which is useful data and roughly zero pitas.

The words on top swing things even harder. I broke down how one title took a video to 651K while the same footage under other titles stalled. My rule: test organically first, small and cheap, beat your own winner, then put money behind the version that won twice.

The agency job now includes getting found

An agency used to stop at the ad. Your buyer does not. They see the reel, then they go looking, on Instagram, on Google, and more and more in ChatGPT. The buyer journey flipped, and if the answer they get back names your competitor, your ad did its job for somebody else.

So before I shoot, I want to know what the machines already say about you. With Hayden Bond at Plate Lunch Collective, I run an AI answer report: what ChatGPT, Gemini, Perplexity and Google say about your brand, your competitors and your category. Sometimes it is flattering. Once, AI had a manufacturer mixed up with a candle company. Either way, the brief changes, and so does what we make. Short form does the selling, while long form answers the questions a buyer types later, which is how video gets you named in AI search. Captions carry more weight than people think now that Instagram is a search engine. And nobody can promise you the AI answer. The people who promise it are selling something else.

The brand still owns the risk

The agency also kept you out of trouble. Hire a creator directly and that job lands on your desk.

Canada's Competition Bureau has sent letters to nearly a hundred brands and agencies about influencer disclosure, and it has said businesses can be held liable for misleading creator posts. Ad Standards publishes the disclosure guidelines most of us work from.

Regulated categories raise the stakes. A hook that wins a trial can still be illegal in the province it played in. Every alcohol title I write gets checked against the province and platform rules, and cannabis against the tighter set. A title that breaks the rules can cost you the whole page, however well it tested.

What it costs, and what you should get for it

My UGC starts at $500 a video. A long form product video runs CAD $1,800 for Canadian brands. Neither number makes me an agency. What makes the difference is what sits around the file: the brief, the versions, the test, the report.

If the price covers one video and nothing else, you bought a video. That can be the right buy. Just do not expect it to behave like an agency, and decide up front whether you want an agency, a marketplace, or a direct hire, because each one hands you a different share of the work.

Red flags

  • One video, one caption, no test plan.
  • Reports views and stops there.
  • Cannot tell you who the trial audience was or where they live.
  • Calls a winner off one round.
  • Asks for your script before asking about your buyer.
  • Has never heard of disclosure rules in your province.
  • Would not use your product, and the video shows it.
  • Will not show you the losers.

What good looks like

A straight read on fit before you pay, including the version where the answer is "this is not for you." A real kickoff: who the buyer is, what they are missing, which number we are moving. Several versions tested cold, iterated, and retested before a dollar of paid. A report that leads with your goal and includes what did not land. Sends watched as closely as likes, because the sale often happens in a DM you will never see. Content built to sell in the feed and get found afterward.

And a creator who turns down products they would not use, because the audience hears it when you fake it. Authenticity is the mechanism, and it is a big part of why UGC works when it works. If you want the full diligence list, I published how to vet a creator portfolio before anybody pays for one.

Questions to ask before a creator becomes your agency

  1. What goal are we measuring against, and what number would tell us it failed?
  2. How many versions will you test, and will I see the losers?
  3. Who did the trial audience turn out to be, and does that match my buyer?
  4. What does AI say about my brand and my competitors right now?
  5. Who checks disclosure and category rules before anything posts?
  6. What would make you turn down my product?

I answer all six before anyone signs anything. You can see the work here, and if you want to know whether a creator should be doing your agency job, start a conversation.

Forbes has it right. Creators are becoming the creative agency. The camera was never the agency, though. The thinking was.


Ben Puzzuoli

Content Creator