Grocery spending is down 5.3 percent at six months. Households above $125,000 fall 8.2 percent. Savory snacks fall 10.1 percent. Limited service restaurants fall 8.0 percent. Yogurt shows the clearest increase. The effects persist through the first year with attenuation after month six.

There is a small forensic detail worth having. The working paper, circulated in January 2025, reported 6 percent, 9 percent for the high income group, 11 percent for snacks, and $416 a year. Every one of those figures came down in peer review. The published numbers are the shaved ones, and the thread is quoting the published set. Somebody did their job.

Gallup, surveying 5,065 adults between May 28 and June 5 of this year, puts current use for weight loss at 11 percent. That was 3 percent in 2024 and 8 percent in 2025. Fifteen percent have ever used. KFF, asking about any indication, gets 12 percent current and 18 percent ever, rising to 22 percent among adults 50 to 64 and falling to 9 percent at 65 and over. Circana reports 23 percent of households contain a user.

One adult in nine, roughly one household in five. That is enough.

The larger numbers in circulation are forecasts wearing the clothes of measurements. Circana's 35 percent of food and beverage units by 2030 is a projection. Morgan Stanley's 55 million users by 2035 is a projection, and its estimate that national caloric intake falls 1.6 percent by then is a projection built on the first one. FTI Consulting's $95 billion of packaged food "at risk" by 2030 is a projection derived from a survey of 1,007 people. None of these are measurements of anything that has happened.

A 1.6 percent move in national calories sounds like nothing. It is not distributed like nothing. The losses land in savory snacks, sweets and limited service restaurants, which is where the margin is, and they land hardest in households above $125,000, which is where the premium pricing is. A category can lose a tenth of its volume from the most profitable end of its customer base and still look fine in an aggregate calorie statistic.

You can watch the industry behave as if it believes this. PepsiCo's Ramon Laguarta fielded a question on the April 2025 earnings call and answered with smaller portions, multipacks, and more fiber and hydration. Nobody reformulates a portfolio over a rounding error.

Conagra closed its fiscal year on May 31 with net sales of $11.3 billion, down 2.9 percent, and a net loss of $1.9 billion. Its executives attribute this to affordability and consumer sentiment. They do not mention GLP-1s. Circana finds restaurant spending among users up 0.9 percent after a year, with casual dining up 4.1 percent, which points the opposite direction from the Cornell group's 8 percent decline in limited service. That conflict is unresolved and I am not going to pretend otherwise.

The packaged food industry is having a bad few years for at least four reasons, and this is one of them. Nobody is saying which.

It reverses. About a third of the paper's households discontinued during the study. On discontinuation the spending effects fade, and the baskets come back slightly less healthy than they were at baseline. The paper says this plainly. The thread does not.

Most people stop. Prime Therapeutics followed 33,607 non diabetic patients starting Wegovy or Zepbound. One year persistence has improved a great deal, from 33.2 percent in 2021 to 60.9 percent for the first half of 2024, and 64.8 percent on tirzepatide. It is still the case that something between a third and two thirds of people are off the drug within a year. A Cleveland Clinic cohort of 7,938 who stopped between three and twelve months found the obesity group had lost 8.4 percent and regained 0.5 percent by one year, with 45 percent holding or still losing, 27 percent switching agents, and 20 percent restarting. Real world weight loss runs just under 9 percent at a year against 15 to 21 percent in the trials.

The basket barely changes shape. Sørensen and colleagues in JAMA Network Open tracked 293 Danish initiators against 884 controls across roughly two million scanned purchases. Per 100 grams the basket lost 2.1 calories and 0.6 grams of sugar and gained 0.3 grams of protein. People are buying less of the same food, not different food.

The word "eating" is doing more work than the evidence can carry.