Marketing + Media Alliance · Ritz-Carlton Bacara, Santa Barbara · July 19–21, 2026
Twenty-one sessions, delivered by the CMOs and analysts who did the work rather than by panelists with opinions. Five findings are now proven well enough to change how a company plans and spends. And every lab arrived at the same second finding: the method worked, the organization was the bottleneck.
The five proven claims
Each of these was run as a real experiment, in market, by a brand that carried the risk. Each one is presented here by the person who ran it, with the proof status stated plainly – including where the evidence stops.
Claim 1. Brand is a performance channel, and most mixes are set wrong
Leading with the offer too early, and too often, measurably damaged long-term results.
Favorable prospects joined at twice the rate. Favorable existing customers were twice as likely to add lines and upgrade devices. A balanced brand-then-offer sequence transacted at twice the rate of a 100% offer schedule and cut cost per incremental conversion by a third. Leading with offer too early and too often measurably damaged long-term results.
Hause refused the brand-versus-performance framing outright. All ads should perform. They do different jobs. Brand generates demand and offer captures it.
Proof status
Measured in market. The same-budget reallocation upside is modeled under conservative assumptions, not yet measured. AT&T is now implementing live.
Claim 2. A master brand’s revenue contribution can be measured at the person level
Parent and product brands complement each other. They do not cannibalize.
Awareness of Intuit raised QuickBooks conversion. Knowing that Intuit is QuickBooks’ parent raised it further. The effect was largest among prospects with no existing QuickBooks relationship, so parent and product brands complement rather than cannibalize each other.
The perception sequence mattered: trust, security and product quality first, then affinity, then AI innovation and human expertise. TransUnion quantified what a single point of brand equity is worth in revenue. Chu was direct that internal attribution modeling had failed at this question.
Proof status
Measured, person-level linkage. Single brand, four-month window.
Claim 3. Movable Middles beat demographic reach, and the high end beats the low end
Campbell’s tested its own hypothesis and was wrong. The high Movable Middles delivered 3x dollars per household.
The original study returned 5x ROAS and 5x household penetration against campaign average. In application, Campbell’s tested its own hypothesis and was wrong: low Movable Middles, at 20% to 50% of category purchases, converted households at 2x. High Movable Middles, at 50% to 80%, converted at 3x and delivered 3x dollars per household, while the low group performed at roughly general population.
Timing spend to high-propensity windows beat early-season loading. Where the framework underperformed, on Goldfish and Chunky, the cause was creative: a single core message rather than varied occasions and benefits. The approach is now extending into retail media and across 16 leadership brands.
Proof status
In production, multi-brand, variance explained.
Claim 4. AI personalization is the largest available effectiveness gain, and it is an operating model change
AI cannot optimize choices it was never given. Creative variation is the binding constraint.
Average lift across the portfolio is about 160%. Indeed Germany returned a 126% increase in account creates, in a market where people rarely change jobs. AT&T remains the ceiling at over 200%. Uber sat near 20% to 29% with creative that was all shades of green. Choice Hotels landed near 24% on an existing media buy with 27 variants. Even the weakest result was about 27%.
Briggs isolated three drivers. Creative variation is the binding constraint, because AI cannot optimize choices it was never given. KPI alignment is second, because a proxy metric will be gamed. Signal density is third, with roughly 15 conversions per creative version as the sweet spot on an inverted U.
Graf was the most useful voice of the Summit on why this is hard. She started in Germany deliberately, to escape US scrutiny. The real output was not the lift. It was one team, one learning agenda, shared audiences and connected signals across brand, growth, product, legal, procurement, agency and country teams that previously ran on competing OKRs and disconnected budgets.
Proof status
Measured repeatedly across roughly 29 experiments. Wide variance, causes now understood.
Claim 5. Creative can be scored and allocated with media-grade discipline
Create nothing new. Change no media. Reallocate to high-scoring assets and conversions double.
The model predicts e-commerce conversion at 81% accuracy before a media dollar is spent. Assets following the 23 guidelines convert at 4x. Half of Kroger’s media spend was running low-converting creative, and that creative was also buying more expensive impressions: a $25 CPA against $11 on Meta.
Reallocating to high-scoring assets – creating nothing new and changing no media – doubles conversions. Vizon also proved brand purpose is the performance driver: real people in relatable food moments beat animated characters shopping aisles, present in top-performing Meta creative 87% of the time against 38% without.
Proof status
Trained on 2025, validated forward on Q1 2026. One retailer, two platforms. Scaling to more audiences and platforms next.
The finding underneath all of them
The measurement problem is largely solved. The operating model is the bottleneck.
Seven labs, independently, hit the same wall after the method worked.
AT&T
Marketers bought in immediately. The colleagues on weekly and monthly cycles did not. The win was reframing the conversation with finance and sales around opportunity cost.
Bealls
Leana Less, the first CMO in the company’s 100-year history across 650 stores in 23 states, inherited what she called a toxic puddle of first-party data. Project ROMI forced it into a first-party data lake and cut the brand health tracker to four drivers. Promoters spent 10x detractors; high-consideration guests 5x. Separating the weather effect from the marketing effect is what unlocked a double-digit increase in marketing investment at a company that had spent a century trying to take money out of marketing.
Mazda
Brad Audet, Chief Marketing Officer, Mazda North American Operations, found every team carrying its own version of the truth. The byproduct of Project ROMI was a single source of truth. He puts roughly $100M of profitability against each point of consideration growth. The CX-5 launch read as a home run in agency and internal reporting, and flat between test and control. His line to take to members: buy people, not impressions.
Indeed, Choice Hotels, Kroger
Budgets, creative supply and unscored creative, respectively.
iOPEX
Nagarajan Chakravarthy, Chief Digital Officer, on agents needing curated knowledge and context rather than data volume, and on plumbing mattering more than model choice. He reframed value from time saved to velocity: 35 test hypotheses moved from a quarter to about three weeks.
Monks
Wesley ter Haar, Chief AI and Revenue Officer, S4 Capital Group, on the bottleneck moving rather than disappearing. Solve content production and the constraint becomes performance scoring and brand and legal sign-off. His sharpest observation: heavier users of the technology are less bullish about its creative ceiling.
Then two studies measured it directly
MarCaps: only 23% of companies agree on why marketing exists
Rebecca Messina, Partner, with research from Dr. Omar Rodríguez-Vilá of the Goizueta Business School at Emory University, across 843 firms. One single-brand, single-market company produced 28 different answers to the question of marketing’s job. 61% claim a shared way of marketing, 42% have it defined, and fewer live it.
MMA and BCG: vision is running ahead of everything that has to deliver it
Janet Balis, Managing Director and Partner, BCG, with 60 major CMOs. CMO vision self-rated at 6 out of 10, running ahead of technology, agency, governance, talent and workflow. Cross-functional at-scale AI initiatives rose 6 points to 21%. Agentic AI is expected to be 27% of the marketing team within three years.
The session we should be proudest of is the one where the answer was no
The model collapses to the statistical middle and to consensus, which is exactly what makes it fast and cheap, and exactly what prevents breakthrough.
Synthetic personas built on Ipsos work were enthusiastic about the output. The output was every trope ever written about breaking barriers. Intuit’s executive creative director tried to work with it and stopped.
Lehault named the mechanism. On brand, on brief and category-literate is not the same as effective. The counterexample is instructive: Dave.com went from 6% to nearly 100% AI-generated TikTok creative by leaning into animation, which demands less precision, as did SC Johnson’s Mr. Muscle revival.
Publishing this is worth more to members than another lift number.
Two things announced first from the stage
Project EG, with the Advertising Research Foundation and the Marketing Science Institute
Scott McDonald, Ph.D., President and Chief Executive Officer, ARF.
Roughly 250 empirical generalizations, each resting on meta-analyses of 50 to 100 peer-reviewed studies, translated into practitioner language, searchable, with a conversational layer. The private alpha opened to the room.
The live quiz did the persuading. Customer assets carry a coefficient of 0.72, against 0.33 for brand assets and 0.04 for advertising, and 75% of promotional lift is brand switching. A simulation put the cost of getting one of these wrong, at a $10B company with $500M in spend, in the hundreds of millions.
A predictive lifetime value lab with AlixPartners
Lauren Beckstedt, Global Chief Marketing Officer, Brunswick Corporation, and Patrick Anglin, Global Co-Leader, Growth Practice Group and Partner and Managing Director, AlixPartners.
Freedom Boat Club had driven lead-to-sale conversion from 3.5% to about 8%, and was out of runway. An individual-level survival model for duration, multiplied by a value model including cost to serve and boat depreciation, rank-ordered a 100,000-customer holdout correctly. Top deciles carry $12,000 to $13,000 of lifetime value against about $5,000.
Conversion rate optimizes for the easiest line to cross rather than the highest-value relationship.
Lauren Beckstedt, Brunswick Corporation
Proof status
Backtested and rank-order validated. The 20% acquisition LTV lift at flat spend is an estimate awaiting an in-market test.
The rest of the record
Also proven, argued and announced in the room
Twenty full sessions were recorded. These are the rest of them.
The room
Three days in Santa Barbara
Closed-door, no press, and built around peer conversation rather than stage performance. Select any frame to open it full screen.
What this room is
Closed-door. No press. The people who ran the work.
For more than a decade, the Marketing + Media Alliance has run multi-year, multi-brand experiments through its Think Tanks and Labs to validate what actually works in marketing. Not panels. Not opinions. Repeatable, financially defensible frameworks, built across dozens of brands. The Summit is the one time each year that the body of work is unveiled in a single room, by the CMOs who validated – or invalidated – the thesis.
Playlist
Join us in 2027
In marketing, value accrues to whoever moves first on what’s proven.
Once a framework becomes common knowledge, the edge is gone. The CMOs who were in this room are already operating on validated work that the rest of the market will not catch up to for years.
The 2027 CMO + CEO Summit returns to the Ritz-Carlton Bacara in Santa Barbara. Attendance is by invitation.