Search for GLP-1 before and after and the results are photographs. The trials that produced the approvals do not deal in photographs; they deal in tables, and the tables show something a photograph structurally cannot — how many people ended up where.
This page reports those tables. Everything in it comes from the efficacy sections of two FDA labels, and no user photographs, testimonials or reconstructed cases appear anywhere on it. What the drugs do biologically is on how GLP-1 works; what they cost is on the cost hub; what happens when people stop is on stopping GLP-1. Nothing here predicts an outcome for any individual, which is a prescriber's territory and not a publisher's, as our editorial standards set out.
The distribution, not the mean
Here is the semaglutide record at the approved weight-management dose, from the Wegovy label's Study 2 — adults with obesity or overweight with a comorbidity, without type 2 diabetes, at week 68.
| Outcome at week 68 | Placebo (N=655) | Semaglutide 2.4 mg (N=1,306) |
|---|---|---|
| Mean weight change | −2.4% | −14.9% |
| Lost ≥5% | 31.1% | 83.5% |
| Lost ≥10% | 12.0% | 66.1% |
| Lost ≥15% | 4.8% | 47.9% |
| Lost ≥20% | 1.7% | 30.2% |
Read the last row first. The outcome that produces a striking photograph — losing a fifth of body weight — happened to 30.2% of patients on the drug. It is a real outcome, it is roughly eighteen times the placebo rate, and it is what happened to slightly under a third of the people who took it for sixteen months.
Read the placebo column second. 1.7% of the placebo group also lost at least 20%, and 31.1% lost at least 5%. A convincing before-and-after exists without the drug, in a minority of people. That is the entire reason these trials have control arms, and it is the fact most absent from the genre.
Tirzepatide, where the top of the distribution is fatter
Zepbound's Study 1, at week 72, in a comparable population:
| Outcome at week 72 | Placebo (N=643) | 5 mg (N=630) | 10 mg (N=636) | 15 mg (N=630) |
|---|---|---|---|---|
| Mean weight change | −3.1% | −15.0% | −19.5% | −20.9% |
| Lost ≥5% | 34.5% | 85.1% | 88.9% | 90.9% |
| Lost ≥10% | 18.8% | 68.5% | 78.1% | 83.5% |
| Lost ≥15% | 8.8% | 48.0% | 66.6% | 70.6% |
| Lost ≥20% | 3.1% | 30.0% | 50.1% | 56.7% |
At the highest dose the picture inverts relative to semaglutide: a majority — 56.7% — reached a 20% loss. But the same table carries the other half of the honesty. 9.1% of patients on 15 mg did not reach a 5% loss, and 29.4% did not reach 15%. A drug with a −20.9% mean still leaves roughly one person in eleven below the threshold most trials treat as the minimum meaningful response.
The two labels are not a head-to-head comparison. They are different trials, different populations, different durations, and the difference between them should not be read as a ranking; the molecule-level comparison this site does make is on tirzepatide vs semaglutide.
Three things the tables show that the photographs cannot
1. Diabetes changes the outcome, and both molecules agree
| Without type 2 diabetes | With type 2 diabetes | |
|---|---|---|
| Semaglutide 2.4 mg, mean | −14.9% | −9.6% |
| Semaglutide 2.4 mg, reached ≥20% | 30.2% | 12.8% |
| Tirzepatide 15 mg, mean | −20.9% | −14.7% |
| Tirzepatide 15 mg, reached ≥20% | 56.7% | 30.8% |
Same drug, same dose, same duration, same endpoint: the proportion reaching the top-end outcome roughly halves in patients who have type 2 diabetes. That this appears across two manufacturers, two molecules and four separate trials makes it hard to dismiss as an artefact of any one study. It is also the single most useful thing on this page for anyone interpreting a result someone else posted, because the population is the first thing a photograph omits.
2. Tripling the dose bought 3.3 percentage points
Novo Nordisk's higher-dose arm supplies a rare within-label dose comparison. In Wegovy Study 8 at week 72, the 7.2 mg arm recorded a mean of −18.8% against −15.5% for 2.4 mg — a difference from the lower dose of −3.3 percentage points (95% CI −4.9 to −1.6), with the 5% responder rate barely separating them at 88.7% against 86.6%.
Three times the drug, three points of mean weight change. The dose-response curve in this class flattens, and a reader who assumes the newest and strongest version of a drug is a different order of result is reading the wrong shape.
3. Some trial "befores" are already a diet's "after"
This is the design detail that most changes how a number should be read, and both manufacturers use it.
- Zepbound Study 3 ran a 12-week intensive lifestyle lead-in — dietician instruction eight times, roughly 1,200 to 1,500 kcal a day, 150 minutes of exercise a week — and only the 579 patients who had already lost at least 5% were randomised into the 72-week trial.
- Saxenda Study 3 used the same shape: a low-calorie run-in, and randomisation only of those who had lost at least 5%.
- Wegovy Study 5 randomised only patients who had reached the full 2.4 mg dose during a 20-week run-in, excluding those who discontinued during titration.
Two of the labels say what follows, in their own words. Saxenda's states that because losing at least 5% during the run-in was a condition of continuing, the results may not reflect those expected in the general population. Wegovy's says the same of Study 5, since patients who could not tolerate titration were never randomised.
A baseline in those trials is the weight of someone who has already succeeded at something. The "before" is not the starting point a reader is standing at.
And one design where everyone got the lifestyle programme
Wegovy Study 4 gave intensive behavioural therapy to both arms. It moved both: placebo reached −5.7% against −2.4% in Study 2, and the drug arm reached −16.0% against −14.9%. The placebo arm of that trial lost at least 5% in 47.8% of patients — half of them — with no drug at all.
The timeframe nobody photographs
These are week 68 and week 72 endpoints. Sixteen to seventeen months, with the first several months spent escalating the dose rather than at the dose that produced the number.
Any before-and-after presented at eight weeks, twelve weeks or three months is describing a different and much earlier part of a curve than the figures above. The trials do publish time courses, and the direction is what would be expected — most of the change accumulates over many months rather than arriving early.
What the record does not establish
- What will happen to any individual. A distribution describes a group. The tables give the odds that were observed, not a forecast, and the variables that move them for a particular person are clinical.
- Why the spread is so wide. The labels report who ended where; they do not explain why two people on the same dose for the same duration land twenty points apart.
- That the trial populations resemble the general one. Two labels say explicitly that theirs may not, and the run-in designs above are the reason.
- What happens after the trial ends. Every figure here is an on-treatment result at a fixed week. Withdrawal is a separate literature, on stopping GLP-1.
- Anything about appearance. Weight change is what these trials measured. Body composition, skin and the rest are not in these tables, and the muscle question has its own page at muscle loss.
- That the numbers are complete. The Wegovy label records body weight missing for 7.2% of drug and 11.9% of placebo patients at week 68 in Study 2, imputed from retrieved subjects in the same arm. The tables are analyses, not headcounts.
How this page was built
Section 14 efficacy tables were read in full from two current FDA labels on DailyMed on 2026-09-18: Wegovy, covering injection and tablets, effective 2026-06-18 (Studies 2, 3, 4, 5, 8 and 9), and Zepbound, effective 2026-08-28 (Studies 1, 2 and 3). Saxenda's Study 3 design and its generalisability caveat were read from its label, effective 2026-02-25. Every proportion and mean above is quoted from those tables as printed; the diabetes comparison sets the labels' own figures side by side and involves no recalculation. No photographs, testimonials or individual cases were used, and none appear on this page.