In 2021, 16.9% of CPG earnings calls promised investment “in our brands.” By 2025 it was 10.8% — and the investment vocabulary had been replaced by a discipline story: efficiency, discipline, cost.
486 of 2,592 dated calls say they’ll “invest in” or “invest behind” our brands — 19%. The timeline is interactive: toggle the full investment vocabulary — or switch to the podcast corpus, where this language barely travels.
How I measured this: six documented regex families swept over the full dated harvest. An earlier, narrower pass reported 308 calls; this re-run uses tighter documented definitions and the full 2,592 dated calls, and the “invest in/behind brands” family landed at 486. 2026 is a partial year.
Only 12 of the 2,592 dated calls say “retail media” — half a percent. That’s the finding; it stands on its own.
The promise names the asset — our brands — but half the time never names where the money goes. That vagueness is the finding: on the call, “invest in our brands” is a pledge without a destination.
Twelve calls in a decade said “retail media” — 0.5% of the 2,592 dated calls, 1.7% in 2025. That’s the finding; it needs no comparison. In the practitioner corpus — commerce and ecommerce shows, harvested precisely because retail media is their beat — the phrase appears in 825 of 5,834 transcripts (19.5% in 2025). That number describes that conversation, not podcasts in general.
Within constant sources there’s no 2020–2025 surge: omni-talk ran ~26→30%, unofficial-shopify-podcast sat at 0% every year 2021–2025, taste-radio 0% in 2024–2025. The pooled 0.7%→19.5% rise is harvesting, disclosed as such.
Same companies, both sides: what execs said on the calls, what the filings show. The two barely move together — the promise leaves no trace in the ad budget. Exploratory, not causal.
Same 91 companies in both eras — a within-company shift, not a sample rotation. What fell out: “price increase,” “supply chain,” “innovation.” And AI? About 2% of calls in both eras — basically absent from the CPG earnings vocabulary.
Exploratory: doc-level term mentions, not dollars. The rent gets paid — it just doesn’t get called an investment anymore.
“Given this backdrop and our ongoing commitment to invest in our brands, we now expect adjusted operating margin to contract slightly for the full year.”
“Trade spend takes you from your gross sales to your net sales. And then your marketing spend takes you from your gross margin down to your contribution margin from a brand perspective.”
I keep a corpus of 5,834 unique CPG transcripts — earnings calls, podcasts, interviews, SHA-256 deduped. Separately, I harvested 2,598 earnings-call transcripts from 152 companies, 2016–2026 (2,592 with usable dates), and swept them with six documented keyword families: execs saying they’ll invest “in” or “behind” (our) brands, build or support brands, invest in marketing/advertising, invest in the consumer, or invest in growth/innovation/R&D — plus anyone saying “retail media.” I added two more families on the same documented pattern (say-vs-spend/f7f8_people_efficiency.py): “invest in (our|the|their|its) people” — invest\w*\s+in\s+(our|the|their|its)\s+people\b — and an “efficiency”-mentions family — \befficiency\b (whole word, any mention per call \u2014 not \u201cinvest in\u201d phrasing) — on the same dated-call denominator. “Invest in/behind brands” showed up in 486 dated calls (19%, peaking in 2021 at 16.9%, 10.8% in 2025). An earlier, narrower pass reported 308 calls; this re-run uses tighter documented regex families over the full dated harvest. 2026 is a partial year. “Retail media” showed up in only 12 of the 2,592 dated calls. The practitioner corpus — commerce and ecommerce shows, harvested precisely because retail media is their beat — mentions it in 825 of 5,834 transcripts; that corpus is topic-selected, so cross-corpus shares are not directly comparable.
I scanned the 486 invest-in-brands calls for six tight channel definitions (digital ads, social, TV, trade promotion, shopper marketing, retail media): 240 name no channel; digital led at 22.2%, “retail media” closed at 1.2%.
The efficiency/discipline shift and the “our people” riser come from a balanced panel of 91 companies with calls in both 2019–21 and 2024–25: doc-level share of calls mentioning each term. Exploratory, not causal — term mentions, not dollars. The say-vs-spend correlation (mention frequency vs ad-intensity change, r≈0.1) covers the 26 companies with both call counts and complete ad-intensity series.
Honest limits, and I’ll own all of them. One: keyword matches map where the conversation goes, not proof of where dollars moved. Two: filings don’t split ad spend by channel, so the channel scan is the say side only. Three: 2026 is a partial year. Every quote on this page comes from the conversation (the call or podcast); numbers it’s paired with come from the filing — never the same source for both.