“Invest in our brands” hit 16.9% of calls in 2021, then faded to 10.8%. Mentions of “efficiency” rose from 41.9% to 63.0% over the same window.
Keyword matches track the vocabulary, not the dollars.
Advertising spend rose from $25.7B to $34.5B — same companies, every year, no sample rotation. The efficiency story didn’t shrink the budget.
“We are seeing a 50% decrease in CPMs… and increase in ROAS just by utilizing our Precision Plus tool.”
“We can be more efficient in how we invest in demand.”
Retail media sits next to the buy button, so ROAS runs hot. Whether that demand is created or merely harvested is the open question.
“ROAS is not the right metric — it’s not only flawed, it can lead you in the exact wrong direction. Return on ad spend is a misnomer: it’s really just saying that a sale happens post advertising exposure.”
“If you look at every short-term window, optimizing for that window and converting, you get more and more targeted over time and the KPIs will look great — but you’ll actually shrink your brand.”
Manufacturers are paying rent instead of building brand equity — the old trade spend, relabeled as retail media. Every time incumbents make that trade, the next wave of insurgent brands gets its opening. The opening is generational: in a 2026 survey of 2,151 US consumers, Gen Z gave national brands 24 points less shelf attention than Boomers — “indifference at the top, not hostility at the bottom.”
“We can be more efficient in how we invest in demand.”
“We are seeing a 50% decrease in CPMs… and increase in ROAS just by utilizing our Precision Plus tool.”
“A lot of the times it starts with how do we grow the retail media network’s bottom line, and moving away from how do we win together.”
I swept 2,592 dated earnings calls from 152 CPG companies (2016–2026; the charts use 2017–2025 because 2026 is a partial year). “Efficiency” is a whole-word match; “invest in our brands” is the F1 family from my six-family investment-language sweep. Both series are shares of dated calls per year. Keyword matches track the vocabulary — they don’t prove where dollars moved.
The spend series is a fixed panel: 42 companies with advertising-expense data in every fiscal year 2017–2025, from 10-K FY facts via SEC EDGAR’s XBRL API. Same companies, every year — no sample rotation. The advertising-expense tag is self-reported and definitions vary by filer.
Two honest limits I won’t blur. One: filings don’t split ad spend by channel, so “most of the new money went to retail media” is my inference — supported by practitioner language, not by a filing line item. Execs almost never say “retail media” on earnings calls (12 of 2,592); the channel’s efficiency pitch is documented in practitioner transcripts. Two: the practitioner corpus is topic-selected (commerce and retail shows), so its enthusiasm isn’t representative of all marketers — I don’t compare its shares against the company-selected earnings calls. The First Insight survey is vendor-authored commercial research; its methodology (sample, ages, categories) is disclosed in the report.
Every quote on this page comes from the conversation (the call, the stage, the podcast); every number comes from the official filing. Never the same source for both.