Scoring creative, and billing without timesheets.
Watch the session
The setup
Every media platform an agency works with can now prove what its own contribution was worth. Reach, targeting, frequency, placement, spend efficiency: all of it is measured, optimized quarterly, and defended in a deck with numbers on it.
The creative work cannot do that. When a client asks whether the idea is the thing that moved the business, the honest answer inside most agencies is a creative director saying trust me.
That asymmetry is not neutral. It decides which line item gets cut first when a budget tightens, and it decides who is in the room when the strategy gets set. FIG's argument in this session is that the creative industry did this to itself by declining to build the data, and that a shop of any size can go build it now.
The session is a walk through the tool they built to close that gap, one campaign it produced, and the commercial model it unlocked on the other side. The second half is the part worth the most to an independent, because FIG has not run a timesheet in seven years and this is what replaced it.
Who was on stage, and why they're worth listening to
Part One — The gap FIG went looking for
Mark's origin story starts with an admission that the agency was chasing something it could not define.
"When we started the agency, we were talking about it. We're like, well, how? What is creative data? So we've always kind of been looking for it. And seven years ago, we found it."
What they found was already sitting in the building. FIG had a strategic map it used with clients to explain how to tell a story and how to argue for a great spot over a good one. The team started whiteboarding scenarios against it, one at a time, by hand.
"Very manual at first, and then year by year, it was like, well, now that it's piling up a bit, this is actually created data. It's not borrowed data, like consumer data. It's actually creative. Then we started automating it."
That distinction between borrowed and owned is the whole thesis, and Judith drew the line under it. The industry has an enormous amount of information about creative. Almost none of it is about the creative.
"What's missing is actual data about the creative, not what do people think about the creative. What is its performance? Who's clicking on it, watching it, doing all of those good things? That's great information that we should all seize. But actual information about what is in a story, what kind of story it is, what signals are in that story is completely missing in the advertising industry."
Her reference point for what the alternative looks like is not another agency.
"It exists in platforms like Netflix with great sophistication. So think about what Netflix did in the taxonomy of entertainment, and then start to apply that to advertising."
Judith put the commercial stakes on a chart, citing Nielsen work she says was first done around 2006 and repeated on a roughly 10-year cycle, quantifying how much of a sales effect is attributable to creative versus media, targeting, reach and brand. Her reading of the trend is the reason FIG built anything at all. (The study attribution and the percentages that follow are as stated on the recording, and are not independently verified.)
"Over the years, creative has always been the most important thing. But as all of the great media companies which we work with and we adore have more and more sophisticated data platforms, they've been able to optimize the success of their element of the story more and more, and creative has got weaker and weaker, in our opinion."
Mark named the number he reads off that chart and named who he blames for it.
"That's a pretty significant jump in 20 years, from 15% to 30 now, and that's ad tech is really what I think that is more than anything."
He then declined the conspiracy reading of it, which is what makes the point land.
"At one point, it's you could look at it very cynically, like they're up to something and the robots and all this shit taking over the world. Or you could say they're applying themselves and they use optimization to do things better, reach more people, tell the right story at the right time. So there's positive stuff there. Our biggest insight really into it is the creative industry, creative departments just took a pass. We didn't get involved with it."
And then the sentence that should be pinned above every independent's strategy department:
"When I look at that ad tech thing, that blue line that keeps growing, those are all really good companies. Great engineering, great tech. There's not a creative person within a mile of any of those companies. They're not making decisions based on storytelling. They're making decisions based on technology. It's not their fault. I put it on creatives. We have to jump in."
The trigger was a holding-company announcement pairing creative with ad tech data, which Mark thought was a smart move badly explained.
"It basically pissed me off. It was great. It's a really smart move for them as a network. It really is. But what pissed me off is the way everyone was talking about it. It's like, oh, we connect these two things, it's done, we have the whole loop, and there's an incredibly big piece missing. And they're a creative company. They should have been the one saying that now creative's going to be the next thing. Nobody's even talking about the fact that it's not there."
His conclusion generalizes past his own product, and it is the most portable thing he says all hour.
"Any company, certainly agencies, are going to need their own proprietary data. That's the only way you're going to be able to use AI in an effective way. Otherwise, we're all going to be using the same thing. What data do you have that you could train AI on and offer something different?"
Part Two — The method, and what it did to a juice brand
Judith walked the mechanics in order, and the order matters because each step is a decision an agency of any size already makes on instinct.
Step one: deconstruct the asset. Any format goes in. 30-second spot, vertical, social, digital, print, out of home. The system separates the images, the text and the sound, then applies models to determine what the asset communicates.
Step two: score it against the four drivers of saliency. Judith defined the goal first, in shopper terms rather than agency terms.
"Saliency being the degree to which a brand comes top of mind when I'm in front of a shelf."
The four drivers are the whole scoring frame: how different, how emotional, how meaningful, and how well branded.
Step three: run it against a competitive set, not the world. This is where the method separates from generic effectiveness scoring. The question is never whether an ad is good. It is whether an ad is different from the specific ads a shopper is choosing between.
"When I'm at shelf and I'm thinking, shall I buy some orange juice, I'm thinking, what kind of orange juice? And so it's really important if I'm thinking different. How different is this story? How different is it to my closest competitor?"
Step four: extend it back through time. FIG pulls up to five years of category history so the answer is a trajectory rather than a snapshot.
Mark's addition is the part that makes the economics work, and he was careful not to oversell the novelty.
"This looks like data analytics, right? Or if you guys are familiar with System1, a lot of people use that. They use the same kind, like it's emotional, different. This isn't groundbreaking. What is groundbreaking about it is we don't have to speak to a person about this. All we do is look at the actual creative. So there's no panels."
And the definition that follows from it:
"So much of what we think is creative testing, it's just focus groups. It's asking people, do you like this. Nothing wrong with that, but that's consumer data. It's not creative data."
He was equally clear that System1 is a collaborator rather than a casualty: "we work with them all the time. We just do something different." On one Publix engagement, the client was running System1 in parallel and FIG matched results against it.
Before any brief gets written, FIG plots the category on what they call a story map, which Judith describes as the genres of advertising. Four of them:
The diagnostic value is in the clustering. Categories usually pile into one genre, which tells you where the empty space is before anyone opens a laptop.
"That's the first conversation we have with a client. Like, well, you're all up here, and you're not spending enough money. So what are you going to do about it? Are you going to continue or are you going to change?"
Mark's addition is that the placement used to be a judgment call and is now arithmetic, and the resolution is finer than a dot on a chart.
"It started, like I said, we put the dots wherever we thought. We just used our brains. Now it's all mathematical. It appears like this one kind of story. It's actually they just spend more time on consumer than on product. So we could tell you, in the past five years, Tropicana, you spent 45% of your time on a consumer story versus your product story."
And what that buys the creative brief:
"We're not shooting in the dark, saying, well, should we tell a benefit story or should we tell a brand ID story. We're briefing that story straight away."
FIG became agency of record for Tropicana roughly a year before this session, and the resulting platform, Give Life Some Juice, had launched about two weeks earlier. Judith walked the diagnosis that produced it.
They built a juice competitive set, set the range at 2021 to 2024, deduped the creative and mapped unique assets against spend. The first finding needed no scoring at all.
"You can immediately see they're the highest spender in the category with the fewest assets. That either means their assets are amazing and they're putting the money behind it, or it means they don't have as many assets as they should to create the breakthrough that they're looking for."
Then the four drivers. Tropicana scored low on differentiation and high on meaningfulness, which Judith translated without mercy: "I can't remember them," and "you meet my needs because you're providing me orange juice." Breaking meaningfulness down further produced the sharper finding: every brand in the category was communicating functional benefits and none was communicating anything emotional.
The last cut is the one that named the brief. FIG scores tone across all categories of advertising on a scale Judith calls happy clappy at one end and sad and somber at the other.
"Juice, which should be happy clappy as citrus is peppy, is actually pretty miserable in how it communicates."
Three findings, one sentence, and it is the closest thing to a creative brief in the session:
"So we know we're not differentiated. We're pretty meaningful, but not about emotion, and we are a miserable brand when we should be a peppy brand."
The pitch answer, Give Life Some Juice, drew the objection every distinctive idea draws. It did not look like the category, and it had a sloth in it.
"They were like, but it's got animals in it. We're like, yeah, because the more animals you have, the more different you are."
The response was not an argument. It was a score, produced fast enough to use in the room.
"So we're doing this just with concepts. We're doing this within 20 minutes of feeding it into the machine."
Mark's line on why that speed exists:
"Because we don't need a panel. We don't have to get a group of people around the country. We just look at the creative, and we get the same results. We could just do it quick and cheap."
FIG then built an AI prototype of the idea, scored the prototype, and only then went into production. Mark's framing of that step is the most defensible use of generative tools anybody put on a Cannes stage this week.
"The prototyping was an important step. That's a really good use of AI. Like, hey, how close could we get the idea, test it in real world. So then they went all in on production with confidence."
After launch, every asset in the multi-channel rollout was scored individually. One did not meet the brief and was changed. Two weeks in, FIG's projection has the brand moving from third to second in category differentiation. (Projection, not result. Flagged.)
Judith's summary of why any of this exists closes the loop back to the opening:
"What we care about is making the best work and making sure that work goes into the world. So all of this is in service of those great stories."
The same instrument that sizes a competitive set and grades a concept also decides how the agency gets paid, and that came out late. The most valuable 90 seconds of the session came from an audience question near the end. A questioner Doug identified as Charlie asked whether the platform has changed the dynamic with clients over time, and whether it is helping FIG get better work into the world. He asked for a genuine answer. Judith gave one, and it went somewhere the question did not.
She started with the competitive problem every shop in the room has.
"One of the big challenges for all of us is you're not distinct just by being an independent agency. You have to be distinct in what you do and how you do it."
Then the operating consequence, delivered as an aside:
"As we've moved, we haven't done timesheets for seven years, everyone's very happy about that. As we've moved to more a value-based way of pricing and an output way of pricing, we can now say we want to increase our differentiation scores by five points. You can pay us against that."
Read that twice, because the mechanism is doing more work than the sentence suggests.
Value-based pricing fails in most independent agencies for one reason. The value cannot be specified in advance in a form both parties will sign. Revenue lift is contaminated by pricing, distribution, seasonality and competitor behavior, so a client will not underwrite it. Brand tracker movement is slow, expensive and arrives after the invoice. That is why shops that talk about value pricing keep quietly running hours behind the scenes.
FIG's answer is to price against a metric they own the measurement of. Differentiation score is produced by their own instrument, on their own asset base, against a competitive set defined at the start of the engagement. It is specific, it is numeric, it moves in weeks rather than quarters, and it is attributable to the work rather than to the market. That is what makes it contractable.
The rest of the model follows from the same asset. Judith described the platform as underpinning every stage of the relationship: positioning and proposition work up front, campaign diagnosis, concept development, post-production optimization, and then ongoing measurement in market. Each of those is a discrete, scopeable deliverable with a defined output. An agency that can name its outputs does not need to count its hours.
One honest limitation. The session does not explain how the scope gets sized, what the payment schedule looks like, or whether there is a floor, a retainer or a bonus structure sitting behind the score. Judith names the unit of exchange and moves on. Anyone building this at their own shop is getting the principle here, not the paperwork.
Judith also added the appended layer for clients who want the harder number:
"One of the things we can do with story data when we're working with the client is then we append it to actual revenue sales data. We will attach it to their media mix model, so we can see what is that asset doing and what is its contribution to sales. That takes away pretty much any argument."
The uncomfortable truth
Mark raised the finding that should worry every agency principal reading this, and he raised it almost in passing.
FIG tracks its own campaigns in market after handoff. What comes back is not what went out.
"Where it's really becoming a breakthrough for us is we could follow it in market, so we don't hand it off to a media company and hope for the best. And what we've come back with is, oh, they're running our campaign. It is not true. What's running is very different than what we thought was going to run."
He immediately refused the villain reading, which makes the point harder rather than softer.
"It's nobody sneaking. It's just optimization happens, and all of a sudden, what's really out in the world is not what you created, and the clients don't really know what's out in the world."
Sit with the second half of that sentence. The agency does not know what is running. The client does not know what is running. The optimization is functioning as designed, and the thing being optimized has drifted from the thing that was approved. Every post-campaign review either side runs is a review of an artefact neither of them has seen in full.
FIG's fix is procedural: assign what Mark calls a creative ID to each asset and track it end to end. Any shop can do a cruder version of that this quarter without building a platform.
The other honest limit came in the same stretch. Asked how the platform handles cultural change across a five-year historical set, Mark's answer was that it explains shifts after the fact rather than predicting them, which is a smaller claim than the room might have expected. Judith reframed toward the long game: salience is built over years, so the historical set is an asset rather than a liability. Neither claimed foresight. Asked about coverage, Mark named the gaps rather than papering over them: experiential is hard because no comparable database exists, and PR has not been attempted.
And on the question every creative department asks, Mark was direct about where the tool does not go.
"We use it to develop a strategy. We use it to develop creative. Not so much. We don't use it in the creative process, really. We're pretty traditional when it comes to that. We go in a room and figure it out."
What you can use
Every session from Indies @ Cannes 2026 — plus every other Blueprint, the tools, and the surveys — opens up with one sign-in.
