The State of Provider-Led Communities in Behavioral Health.

We started Labs because the questions our customers asked us most were questions nobody could answer with published data. How reachable is a typical alumni base at ninety days? What do families actually participate in after discharge? Every program has an opinion. No one has a number. This study exists to put numbers where the opinions are, in the open, under a method we commit to before we see a single response. We will publish whatever the data says, and we will do it every year.

Faris Sweis, Co-Founder & CEO, Cuepri August 2026

Nobody has ever measured the thing every program claims to run.

Most treatment organizations say they run an alumni program, and no published benchmark exists anywhere in behavioral health for community engagement, family participation, referral economics, or the evidence those programs leave on the table. Treatment outcomes are measured. The community around the client is not.

• Q1
  Could you reach your alumni from ninety days ago? Programs know who discharged. Far fewer know who is still reachable at thirty days, ninety, a year. The decay curve has never been published.
• Q2
  What share of your admissions do alumni actually drive? The cheapest admission is the one you already earned, and most organizations cannot name the number. We convert it into dollars per hundred admissions.
• Q3
  How many families are still engaged ninety days after discharge? Trained, engaged families are among the strongest levers in the evidence base, and no one has ever published a family participation number. This study will be the first anywhere.
• Q4
  What does missing between-session evidence cost you? What happens between sessions decides utilization review. We ask how many authorized days were lost in the past ninety for lack of a record.

Six metrics, named in writing, free for anyone to adopt.

A benchmark is only useful if the field can repeat it. So the study does not just report numbers, it names the yardsticks, defines each one in writing, and releases the definitions openly. These are the six the 2026 edition establishes.

An open standard, on purpose. Every definition is free to quote, cite, and reuse, including by people who compete with us. If the field standardizes on these names, the study did its job.

Eight sections, thirty-seven questions, twelve minutes.

We define community the way the evidence does: it begins the day someone commits to treatment, not the day they leave, and it includes clients in care, the alumni before them, and the families beside both. Every metric ships with an open definition anyone may adopt.

Reviewed before we field, published whatever the data says.

Cuepri builds software for provider-led communities, so we have an obvious interest in what this study finds. The only defense against that is method, stated in advance and held to in public. A benchmark that flatters its publisher is worth nothing.

Aggregate only

Nothing published identifies a respondent or an organization. No vendor scorecards, no league tables.

Minimum cell of five

No cut is reported unless at least five organizations sit inside it. Thin segments are disclosed as thin. This is enforced in the code that produces the figures, not remembered at write-up: a smaller cut carries no counts at all, so it cannot be printed by accident.

One response per organization

Deduplicated by email domain. Later responses from the same organization are kept and flagged, never discarded, so exclusions can themselves be disclosed.

Speeders and straight-liners excluded

Elapsed time is measured from the moment the instrument is opened, on our servers, not from anything the browser reports about itself. Excluded responses are stored and counted, not deleted.

"We don't track this" is never a zero

It is excluded from every median and reported separately as its own rate. An organization that does not measure something is a finding about the field, not a hole in the data.

Banded answers

Reported as medians of band midpoints, which is only as precise as the bands allow. Open-ended top bands take midpoints close to their floor, erring low rather than flattering the field.

Reviewed before fielding

External reviewers critique the instrument before a single response is collected. The review panel is named with the instrument lock on September 1, and their notes are printed, including the criticism.

Method published in full

Sampling, instrument, definitions, exclusions, and limitations ship alongside the findings. This section is that method.

No outbound without opt-in

Your responses are never used to target you. Four separate opt-ins, never one box. Sales contact happens only if you ask for it.

Free with attribution

The report and every metric definition are free to quote, cite, and reuse. We want the field to adopt them.