What 26 conversations add up to.
The argument
Across three days and 26 sessions, the same structure kept appearing under different subjects.
An independent agency is handed a scoreboard it did not build. Procurement's scoring matrix. The holding company's technology stack. The platform's black box. The awards shortlist. The billable hour. The client's brand tracker. Each one measures something, and none of them were designed by the people being measured.
The sessions that produced real answers all did the same thing in response. They did not compete harder on the existing number. They changed the number.
Steve Hill built an eight-point creative scale so that 70 North American brands at Kraft Heinz could argue about work using shared language instead of taste. FIG stopped keeping timesheets and started billing against a differentiation score they own the instrument for. Oakley weights sponsorship on performance once, authenticity twice, community three times, which is a deliberate refusal of reach as the governing metric. Kraft Heinz swapped Lions for Effies as the internal scoreboard. Yadira Harrison answers capability gaps with a named coalition rather than a white label. Katy Marshall told a room full of agencies to kill brand awareness as a goal.
And Chaucer Barnes named the mechanism directly, in the first session of the week, without anyone realizing he had supplied the week's thesis:
"It's really about being able to optimize to any single point that you desire. You can make it about margin, you can make it about headcount."
Inside a holding company, that number is chosen for you, quarterly, by someone you have never met. That is the actual difference independence buys — not freedom in the abstract, but authorship of the measurement. The independents who are winning stopped competing on numbers they did not choose.
Most of the rest are still optimizing against someone else's scoreboard and calling it discipline.
Four arguments the week never settled
A synthesis that only found agreement would be a press release. These are the places where the 26 sessions contradict each other, and each one is a decision waiting on the reader's desk.
Killing The Kitchen Sink split three to one. Bill Oberlander, Yadira Harrison and Stacey Gersten all argued that staying small protects the product — Harrison's "lose the sauce" is specifically about senior people no longer doing the work. Chaucer Barnes disagreed: "I don't have a problem with bigger on the face of it. I have a problem with bigger to the extent that it dilutes your proposition."
The rest of the week sided with Barnes more than with the majority.
Zambezi built a media capability in-house from retained earnings, and Liquid I.V. arrived through that owned capability — a direct contradiction of Oberlander's model of outsourcing media wholesale. Britton Upham took enterprise business off holding companies by arguing breadth, not specialization. Heidi Singleton merged six agencies across three countries and kept the work standard intact. Alison Monk's alliance model exists precisely so a small shop can present as a large one.
The honest reading is that smallness is not the variable. Ownership of the instrument is the variable. Oberlander and Zambezi disagree about headcount but agree about control — one rents the capability and keeps the proposition clean, the other builds the capability and keeps the margin. Both know which number they are protecting.
This was the loudest unresolved fight of the week, and it ran across at least five sessions.
Kelly Mertesdorf argued differentiation from Ritson and brand health trackers. Pat Laughlin argued Bob Dylan going electric, and Doug Zanger's rebuttal — that Dylan had record sales — was left standing on stage. Steve Hill built a scale, then warned that the scale gets weaponized to enforce opinions. FIG built an entire company on scoring creative in 20 minutes without consumer panels. Katy Marshall and Alison Monk disagreed on stage about whether "the winning ad" is the goal or a by-product of a properly instrumented system.
Nobody reconciled it, and the reason is visible in Hill's own caution. Every measurement system on offer was built by someone who benefits from it being adopted. The scale that creates shared language also creates a veto. That is not an argument against measurement. It is an argument for knowing who built your instrument and what it was built to protect.
Beyond The Doomscroll contained the week's sharpest labor disagreement. Josh Archer said AI removes inhumane volume work and gives people their evenings back. Chelsea Kauth said her creative team is working longer and harder than before.
Both are describing the same tool at different points in the value chain, and the billable hour sits underneath both. If a job that took nine weeks now takes three, someone loses two-thirds of the revenue unless the pricing model changes — which is why FIG's abandonment of timesheets keeps looking less like an operating quirk and more like a preview.
The labor question got harder in The Guardrails Of Imagination. Baptiste, from the floor, named a $20 barrier keeping juniors out of the tools. Johnny Rohrbach countered that the delta between a junior and a senior operator has never been smaller. Both cannot be comfortably true — if the delta is smaller, the $20 barrier is the whole ballgame.
And The Kids Are Not Alright supplied what the AI sessions did not: AI absorbed exactly the tasks that used to train juniors, which means the industry has automated its own apprenticeship and has not replaced it.
Chaucer Barnes attacked client-conflict conventions in the first session: "It's not like you work on Coke and they tell you the recipe. Everybody's plans is everybody's plans." Nobody on his panel picked it up.
Three sessions later, Tina Fegent described category-experience screening in UK procurement colliding with those same conflict rules. Britton Upham described holding-company technology lock-in as the thing that keeps clients captive. And then Mark Figliulo built an entire proposition on the opposite premise — own your data, own the instrument, and price against it.
That is a live disagreement between two independents who never shared a stage, about the single question that determines whether independents can act as a bloc or only as competitors. It is the most consequential unfinished argument of the week.
What was conspicuously missing
Four absences, in ascending order of importance.
Nobody costed AI. Across two dedicated AI sessions and half a dozen more that touched it, no one put a number on what any of it costs to run at agency scale, or what it does to a rate card. Melanie Ryan asked the panel directly about bias and brand-safety auditing and got no answer.
Nobody defined the guardrails they were demanding. Audience Understanding produced the same absence in a different form — an agency accused ad tech of expanding audiences silently, a publisher agreed, and confirmed there is no industry standard. Nobody would say what acceptable expansion looks like.
The client's side went largely unrepresented, except where it was devastating. Confessions Of A CMO was the only session built on original research, and John Barger's admission — "it's so easy to blame the agency, there's a scapegoat, I think it's a cop out" — is the exact mirror of Chuck McBride's "so you lied." Both sides of the table confessed to the same evasion, on different days, to different rooms, and neither heard the other.
The work itself. This is the one that matters, and it did not come from the synthesis. It came from the floor at the closing session, where a speaker charged that Cannes 2026 talked about AI and never about the work. Katie Walley-Wiegert named the same gap more precisely: the missing register is the in-between one, where agencies talk about the fight they lost with a client rather than the case study they won.
That charge is testable, and it should be tested rather than repeated. Counting how many of the 26 sessions led with AI versus led with the work would convert a good line into a finding. Until then it stands as the most uncomfortable thing anyone said all week, delivered about the week itself, in the room that was supposed to be summing it up.
The one-sentence version
Independence is not the absence of a boss. It is the authorship of the scoreboard — and most independent agencies are still being scored on a number they did not choose, by people who will never explain how it was calculated.
Where this leads
Four things the sessions collectively point at, none of which any single session delivered.
Every session from Indies @ Cannes 2026 — plus every other Blueprint, the tools, and the surveys — opens up with one sign-in.
