The selection problem
A telehealth program that publishes ten testimonials of patients losing 25–35% of their body weight has not, by doing so, given a prospective patient any information about likely outcomes. The denominator is missing.
If 1,000 patients were treated and the ten most successful are featured, the conditional probability of outcomes in that range — for a new patient drawn from the same intake stream — is unknown. It might be 1%, 5%, or 30%. The testimonials don't say.
What a denominator looks like
A cohort statistic provides the missing denominator. 'Among 1,247 patients who completed at least 26 weeks of therapy with our program in 2025, mean total body weight loss was 14.2% with an interquartile range of 9.8 to 18.6%' is a statement a prospective patient can act on. It also exposes the program to comparison: a competitor publishing 17.1% on a similar cohort is, by inference, doing something the first program is not.
This is what cohort outcomes data is, and what testimonials are not.
Why almost nobody publishes it
Publishing cohort data exposes you to competitive comparison, to skeptical statistical scrutiny, and — if your outcomes are mediocre — to direct negative inference. Almost no program in the compounded-telehealth segment publishes. The Outcomes Transparency pillar in our rubric absorbs this directly: most providers score 12–17 of 20 not because they're producing bad outcomes, but because the data to verify outcomes one way or the other does not exist.
A program that does publish cohort outcomes — even modest ones — scores meaningfully higher on that pillar than a program with strong-looking testimonials and no denominator.