The Campaign Worked

In November 2023, Snoop Dogg told his 129 million followers that he was "giving up smoke." Fifty-two million people watched the reveal video. The campaign generated over 30 million engagements and billions in earned media impressions. Ad Age ranked it eighteenth on its list of the 40 best ads of the year.¹

The client was Solo Stove. The product was a smokeless fire pit.

By every metric the marketing industry uses to evaluate a campaign, this one worked. Awareness, engagement, earned media, creative recognition. The dashboard was green across the board.

In January 2024, Solo Brands replaced its CEO. The interim CFO stated that the campaign "raised brand awareness of Solo Stove to an expanded and new audience of consumers" but "did not lead to the sales lift that we had planned." By the third quarter of 2024, net sales had fallen 14.7% year over year to $94.1 million, with a net loss of $111.5 million. Third-quarter sales dropped another 44% the following year. In April 2025, Solo Brands received a NYSE delisting notice. A 1-for-40 reverse stock split in July lifted the shares back above a dollar. In April 2026 the NYSE suspended the stock anyway, and it moved to the over-the-counter market.²

The campaign worked. The company did not survive it.

Two measurement systems have been running in parallel for sixty years without anyone requiring them to reconcile.

The two dashboards

Every company that spends money on marketing is running two measurement systems simultaneously. One tracks what the marketing department does. The other tracks whether the business survives.

Both systems produce real numbers. Both are internally consistent. A campaign can score perfectly on every metric in the left column and produce zero movement in the right column. There is no mechanism, in most organizations, that requires the two to connect.

Gartner surveyed 378 senior marketing leaders in 2024 and found that only 52% could prove marketing's contribution to business outcomes. Nearly half of the people running marketing could not demonstrate that their work moved the numbers their company uses to determine whether it is succeeding.³

The CFOs already knew. In the same research, 40% of senior marketers identified the CFO as among the top three executives most skeptical of marketing's value.³

The CFO watches the right column. The marketing department reports the left column. The two columns do not converge, and nobody in the organization has the authority or the incentive to force the question.

What happens when you turn it off

The cleanest test of any system is what happens when you remove it.

In 2017, Procter & Gamble, the world's largest advertiser, cut $200 million in digital ad spending. In the quarter they cut the first $100 million, organic sales rose 2%, on net sales of $16.1 billion. Net earnings rose 12%. P&G;'s chief brand officer, Marc Pritchard, said the average dwell time on their mobile news feed ads had been 1.7 seconds. Adweek reported that cutting $200 million in spend actually increased P&G;'s reach by 10%.⁴

The same year, JPMorgan Chase reduced its programmatic ad placements from 400,000 websites to 5,000. A 98.75% reduction. CMO Kristin Lemkau reported little change in the cost of impressions or the visibility of their ads. Performance metrics were unchanged.⁵

At Uber, the head of performance marketing cut two-thirds of a $150 million programmatic budget. No measurable drop in app installs or rider acquisition. The investigation revealed that ad networks had been claiming credit for organic conversions that would have happened regardless. Uber sued its mobile agency and five ad networks for over $70 million in attribution fraud.⁶

In 2012, eBay collaborated with academic researchers to halt its paid search ads. The result, published in Econometrica, one of the top five economics journals: almost all paid search traffic was substituted by free organic channels. Brand-keyword ads had no measurable short-term benefit. Average returns on paid search spending were negative.⁷

Four companies. Four independent experiments. The same finding: a significant portion of what appeared on the marketing dashboard as activity was not generating incremental commercial results. The left column was green. It was measuring itself.

These experiments cluster between 2012 and 2017. The best-known test since then was forced, not chosen. In 2020 the pandemic made Airbnb pull all of its marketing, and its traffic came back to 95% of the 2019 level before the spending resumed.⁷ Few companies run the test on purpose. The answer is uncomfortable, and the system that produced the uncomfortable answer has no incentive to ask the question again.

The elasticity

The academic literature has been quantifying this disconnect for decades. Most of the marketing industry has not read it.

In 2011, Sethuraman, Tellis, and Briesch published a meta-analysis in the Journal of Marketing Research covering 751 short-term and 402 long-term advertising elasticities drawn from 56 studies between 1960 and 2008. The average short-term advertising elasticity was 0.12. A 1% increase in advertising spending produces a 0.12% increase in sales.⁸

In 2025, Korkames, Stanley, and Stremersch published an updated meta-analysis correcting for publication bias and methodological problems in the original studies. The corrected short-term advertising elasticity fell to 0.0008. Statistically indistinguishable from zero. The long-term elasticity was 0.03.⁹

For comparison, Tellis's 1988 meta-analysis of 367 econometric models found a mean price elasticity of −1.76. A 1% change in price moves demand 1.76%.¹⁰

This is a meta-analytic consensus spanning 65 years of econometric studies. Price, distribution, and shelf position explain dramatically more purchase variance than advertising. The measurement system that governs most marketing budgets is oriented toward the smallest lever in the equation.

Where the decision happens

The money goes to what happens before the purchase. The purchase itself happens somewhere else.

The Point of Purchase Advertising International conducted shopper engagement studies in 2012 and 2014 using eye-tracking, EEG, and in-store observation. In grocery, 76% of purchase decisions were made in-store. In mass merchants, 82%. Thirty-four percent of mass merchant shoppers had no list at all.¹¹

In 2025, Vestcom surveyed 2,000 U.S. consumers and found that 72% make unplanned purchases based on in-store messaging. In-store marketing was three times more likely than digital ads to influence a shopper to try a new product.¹²

In 2010, Ataman, Van Heerde, and Mela published a study in the Journal of Marketing Research tracking 70 brands across 25 product categories over five years. They measured the total elasticity of each element in the marketing mix. Distribution had a total elasticity of 0.74. Advertising had 0.13. Getting the product into the store and onto the shelf moved sales nearly six times more than advertising did.¹³

Byron Sharp at the Ehrenberg-Bass Institute documented the mechanism across decades of research. Brands grow through mental availability, the probability of being recalled in a buying situation, and physical availability, the ease of finding and purchasing the product. Advertising contributes to mental availability. But physical availability is the prerequisite. Sharp's formulation is direct: advertising falls on barren ground when it reaches buyers who are not near the firm's sales points.¹⁴

The decision happens at the shelf. Price, placement, availability, and whether the product is there when the shopper is. The marketing dashboard measures what happens before the shopper walks through the door. The business dashboard measures what happens after. Most of the variance is in the after.

The mechanism

The parallel systems persist because every participant in the chain has a rational incentive to maintain them.

The CMO

The average CMO tenure at an S&P; 500 company is 4.1 years. The CEO averages 7.6 years. The CFO averages 4.7. Only the COO has a shorter tenure, and that role is typically a stepping stone to CEO.¹⁵

A 4.1-year window creates a structural incentive. Awareness moves in weeks. Engagement moves in days. Sell-through rates, margin trends, and reorder patterns take years to reveal themselves. A CMO whose next role depends on demonstrable results within four years will optimize for metrics that move within four years. Spencer Stuart found that 62% of departing CMOs leave for similar or bigger roles. Nine percent get promoted to CEO.¹⁵

The incentive is to measure what you can move before you leave.

The agency

The Association of National Advertisers found in 2022 that labor-based fees remain the predominant agency compensation model. Performance incentives tied to business outcomes had been declining for nearly a decade, falling to 41% of advertisers. Media agencies typically earn a commission of 10-20% on media spend. Their revenue increases when the client spends more.¹⁶

An agency whose revenue depends on the volume of spend it manages will not recommend less spend, even when the evidence suggests that less spend produces the same outcome. The compensation design makes the recommendation before the strategist enters the room.

The attribution model

Avinash Kaushik, one of the most cited voices in digital analytics, published a detailed examination of marketing incrementality in 2021. His conclusion, drawn from years of analyses across company types: the true incrementality of marketing is typically between 0% and 25%. Meaning that 75% to 100% of conversions attributed to marketing would have happened anyway, through organic search, direct traffic, word of mouth, or the simple fact that the company exists.¹⁷

Attribution distributes credit. It does not measure causation. Every major attribution model takes the conversions that occurred and allocates them across the marketing channels that touched the customer beforehand. None of them ask whether the conversion would have happened without any marketing at all. The entire system exists to answer the question "which channel deserves credit?" while skipping the prior question: "did marketing cause this?"

The inversion

In October 2025, Les Binet presented data from the IPA Effectiveness Databank spanning 1998 to 2025. The finding was precise: 89% of variations in incremental profit are explained by budget size. Only 11% are explained by ROI efficiency.¹⁸

When Medialab asked 500 senior marketers which mattered more, they answered the other way. Budget: 35%. ROI: 65%.

The entire optimization infrastructure, the A/B tests, the attribution models, the ROAS calculations, the agency performance reviews, is oriented toward the 11% variable. The 89% variable is the budget itself, and nobody optimizes that because the system is designed to justify it, not question it.

Peter Field's companion finding for the IPA: marketing effectiveness declined by about a third in the decade to 2018. The IPA tracked the average number of "very large business effects" reported by campaigns. It fell from approximately 2.0 in 2008 to 1.3 in 2018. The multiplier for creatively awarded campaigns collapsed from 12x to 4x over the same period.¹⁹

The industry got more sophisticated. Measurement became more granular. Dashboards updated in real time. The effectiveness went down.

The origin

The split between marketing measurement and business measurement was designed.

In 1961, Russell Colley published Defining Advertising Goals for Measured Advertising Results, known in the industry as DAGMAR. His proposal was that advertising should be evaluated on communication objectives, awareness, comprehension, conviction, rather than sales. The rationale: too many variables influence sales, so isolating advertising's effect is impractical. Instead, measure what advertising can plausibly control.²⁰

The framework was adopted. It gave the industry permission to measure itself by its own standards. Sixty years later, the permission has become infrastructure. Marketing departments report awareness. Finance departments report revenue. The two sit in different meetings, use different dashboards, and answer to different questions. Nobody is required to reconcile them because the foundational framework said they didn't need to be reconciled.

Adidas discovered the cost of this architecture in 2019. Simon Peel, the company's global media director, revealed that Adidas had been spending 77% of its marketing budget on performance and digital marketing and only 23% on brand building, driven by last-click attribution models. When they ran econometric modeling for the first time, they found that brand activity was actually driving 65% of sales across wholesale, retail, and ecommerce. The attribution model had been systematically misrepresenting what drove purchases for years.²¹

Peel's admission was direct. "We had an understanding it was digital advertising driving ecommerce sales and as a consequence we were over-investing." The company shifted its approach. But the years of misallocation had already happened, guided by a measurement system that was internally consistent, technically sophisticated, and wrong.

The proof cases

Solo Stove is not an outlier. The pattern repeats across industries and scales.

In April 2023, a single sponsored Instagram post from Bud Light to transgender influencer Dylan Mulvaney triggered a consumer boycott. Awareness was enormous. AB InBev lost an estimated $1.4 billion in North American revenue that year. Bud Light dropped from the number one to number three U.S. beer by market share. Multi-million dollar Super Bowl advertising failed to recover the position. As of mid-2024, the effects endured.²²

Quibi raised $1.75 billion and launched with a Super Bowl ad, Hollywood-level content, and massive pre-launch awareness. It shut down after six months. Nobody lacked awareness of Quibi. Nobody wanted the product.²³

Peloton achieved enormous brand awareness and cultural penetration during the pandemic. Instructors became celebrities. Market capitalization reached nearly $50 billion. By early 2026, it had fallen to approximately $1.8 billion, a 97% decline. The brand awareness remained. The business did not.²⁴

In each case, the marketing dashboard was green. The awareness was real. The engagement was real. The measurement system declared success while the business moved in the opposite direction. The two dashboards were producing accurate readings of two different things, and the thing the marketing dashboard measured had no causal relationship with the thing that determined survival.

The belief

I have spent 25 years watching dashboards say one thing and bank accounts say another. Not at one company. Across industries. The pattern does not change because the architecture does not change.

The belief is that the campaign worked. The campaign always worked. The measurement system that evaluates campaigns is designed to confirm that they worked. Impressions were delivered. Awareness rose. Engagement was generated. The creative was recognized. By its own logic, the system never fails.

The question the system does not ask is whether any of it moved the number that determines whether the business exists next year. Revenue. Margin. Velocity at shelf. The reorder. The metrics that sit on a different dashboard, reported by a different team, in a different meeting, using a different language.

The 1961 DAGMAR framework said these didn't need to be the same measurement. Sixty years of infrastructure was built on that permission. And in those sixty years, the academic evidence accumulated: advertising elasticity is 15 to 59 times smaller than price elasticity. Three-quarters of purchase decisions happen in the store, not before it. When major companies cut hundreds of millions in digital spend, the business metrics do not move. The true incrementality of marketing is, in the most generous studies, 25%.

The campaign worked. It always works. The measurement system guarantees it.

The business runs on a different system entirely. And the distance between the two is where the money disappears.