Annual report
No. 01

First annual edition

The 2026 Behavioral Health Patient Acquisition Report

Cost per admission, channel economics, and AI era discovery, measured across $64M+ in managed media and 37,200+ attributed admissions.

Published 14 November 2026 Managed accounts, 2018 to 2026 27 markets, 9 states
$64M+Google and Meta media managed since 2018
37,200+Admissions tracked, same period
27Markets across 9 states
90%+Annual partner retention

Summary

Five findings

  1. Cost per lead cannot be compared between programs. In a worked example with hypothetical inputs, two facilities buying identical media at an identical $200 per lead differ sixfold in what an admission actually costs them.
  2. Blended cost per admission across the measured portfolio is $618. Paid media only cost per admission is $1,600.
  3. The relationship between those two figures caps paid media at roughly 39% of admissions. The majority of admission volume carries no per admission media cost.
  4. At revenue near $31,000 per admission, gross return on acquisition spend is approximately 50:1 blended and 19:1 on paid media alone.
  5. Across 125 measured AI assistant queries, portfolio facilities appeared as a cited source in 38.4% of responses. Almost no operator in this category measures this channel at all.

Every figure in this report, at a glance

FigureWhat it measuresBasis
$618Blended cost per admission, all channelsMeasured, current reporting period
$1,600Cost per admission from paid media aloneMeasured, current reporting period
Near $31,000Average gross revenue per admissionMeasured, portfolio average
39%Implied paid share of admissions, upper boundDerived, $618 ÷ $1,600
61%Admissions from channels with no per admission media costDerived, 1 − 0.386
2.6×Paid cost per admission against blendedDerived, $1,600 ÷ $618
50:1Gross return on acquisition spend, blendedDerived, $31,000 ÷ $618
19:1Gross return on acquisition spend, paid media onlyDerived, $31,000 ÷ $1,600
$1,720Lifetime media per attributed admission, 2018 to dateDerived, $64,000,000 ÷ 37,200
38.4%AI assistant citation rate on the tracked question setMeasured, 48 citations of 125 observations
$64M+Google and Meta media managed since 2018Measured, cumulative
37,200+Admissions tracked since 2018Measured, cumulative
27 markets, 9 statesFootprint of participating programsMeasured, current
90%+Annual partner retentionMeasured, current

Measured figures come from managed advertising accounts and participating program admissions reporting. Derived figures are computed from measured figures only, with the arithmetic shown in the section named. One derivation in section 03 compares a published outside figure against one of these, and it is labelled where it appears. No other derived figure depends on anything published outside this report.

01

Why cost per lead is the wrong metric

Nearly every media plan in behavioral health is sold, budgeted and reported on cost per lead. It is the wrong denominator, and it is the reason most operators cannot tell a good quarter from a bad one.

A lead is an inquiry. An admission is a patient in a bed. Between those two events sit insurance verification, clinical fit, bed availability, family consent, and above all the speed of the admissions team. The conversion rate across that gap is not a property of the media. It is a property of the program, and in practice it varies by an order of magnitude between facilities buying the same inventory at the same price.

Worked example

Two programs, identical cost per lead. Hypothetical inputs.

Program A  $200 CPL, 12% lead to admission
$200 ÷ 0.12 = $1,667 per admission

Program B  $200 CPL, 2% lead to admission
$200 ÷ 0.02 = $10,000 per admission

6.0× spread

The inputs are hypothetical, not portfolio figures. The point is structural. Same channel, same creative, same price per inquiry, and a sixfold difference in the only number that determines whether the program is solvent.

An agency reporting a falling cost per lead while cost per admission rises is not performing, and the operator has no way to see it. For that reason every figure in this report is denominated in admissions.

02

The portfolio behind these benchmarks

The benchmarks in this report are drawn from $64M+ in Google and Meta media and 37,200+ attributed admissions across 27 markets in 9 states, managed from 2018 to the current period. The portfolio totals on the cover are cumulative across the full managed history. They establish the size of the base the unit economics are drawn from. They are not an annual figure, and they should not be multiplied against current period rates. The methodology returns to this, because it is the most common way a report like this gets misread.

Derivation

Lifetime media per attributed admission

$64,000,000 ÷ 37,200 = $1,720

~$1,720 lifetime

Computed on the floor of both published figures, each of which is stated with a plus. The average spans the entire period, including the early years before nonpaid channels existed in these accounts. Current period blended cost per admission is $618. The distance between those two numbers is the subject of the next two sections.

03

What a behavioral health admission costs

$618
Blended cost per admission, all channels, current reporting period
$1,600
Cost per admission from paid media alone, same period
Near $31,000
Average gross revenue per admission, portfolio average across levels of care and payer mix

An operator benchmarking against these should hold two things constant before drawing a conclusion. Admissions counted at admission date against census, not at assessment. And gross charges rather than collected revenue. The methodology states both definitions in full.

Published cost per admission benchmarks in this category

Published cost per admission figures in this category are below, with what each one measures and the sample behind it, so the definitions can be compared before the numbers are.

SourceFigureWhat it measuresSample and period
This report$1,600 paid
$618 blended
Media spend divided by attributed admissions37,200+ admissions and $64M+ media across 27 markets, 2018 to the current period
Webserv$16,608PPC spend divided by attributed admissions, described on their page as the full funnel from click to census$16.5M of PPC spend and 994 attributed admits across their full client base, 2025
Lead to Recovery$600 to $914Cost per admission on organic search at individual client centersSingle campaign figures at named centers. No portfolio total stated
Active Marketing$1,650 to $12,000Total marketing spend divided by enrolled patientsCited as an industry range. No sample stated

One of these is built the same way as the figure above. Webserv's figure and the paid figure in this report are both media spend divided by attributed admissions. Active Marketing's covers total marketing spend, which is a broader measure and is not directly comparable to either. Lead to Recovery's figures are single channel results at individual centers rather than a portfolio measure, so they are not comparable to a blended or paid portfolio figure in either direction.

Derivation

The two like for like figures, compared

$16,608 ÷ $1,600 = 10.4

10.4×

Both inputs are published by their sources and linked above. What the arithmetic does not settle is attribution. This report attributes paid admissions on platform reported conversion reconciled against program census monthly, stated in section 08. Webserv's page does not publish its attribution method at that depth, and its client base is described as mixing in network and out of network programs, which moves cost per admission on its own. The figures are comparable in construction. Whether they are comparable in rigour cannot be established from published material alone, and this report does not claim it.

04

Where treatment center admissions actually come from

Paid media accounts for no more than 39% of admissions in the measured portfolio. The other 61% arrive through channels that carry no per admission media cost. That figure is not surveyed, it is forced by the two published cost figures.

The blended and paid costs per admission are not independent of one another. Their ratio constrains how much of the admission volume can possibly be arriving through paid media.

Derivation

Implied paid share of admissions

blended = (paid CPA × paid share) + (nonpaid CPA × nonpaid share)

Treating nonpaid channels as carrying no per admission media cost

$618 = $1,600 × p
p = 618 ÷ 1,600 = 0.386

39% paid, at most

An upper bound. The calculation does not allocate fixed costs such as retainers, content production and technology across admissions. Allocating them pushes the implied paid share below 39%, not above it.

So roughly 61% of admissions in this portfolio arrive through channels that carry no per admission media cost. Organic search, AI citation, referral relationships, direct demand and returning alumni do not bill per admission once they exist.

Derivation

What paid only volume costs by comparison

$1,600 ÷ $618 = 2.59

2.6×

An operator putting the entire acquisition budget into paid media is buying about 39% of the available volume at 2.6 times the blended cost.

This is not a claim about creative, bidding or account hygiene. It is a statement about channel composition, and it is the most consequential number in the report. Paid media is fast, controllable and expensive. In a mature account it is also the minority of admission volume. A program that has never built the other 61% is paying list price for every single admission it takes, permanently, and will keep doing so no matter how well the campaigns are run.

The corollary matters as much. The 61% is slow to build and cannot be bought in a quarter. That is precisely why it holds its value. It is also why the lifetime average of $1,720 per admission and the current blended figure of $618 differ by so much. The gap is not a bidding improvement. It is eight years of compounding in channels that have no marginal cost.

05

Return on ad spend for a treatment center

Derivation

Gross revenue multiple on acquisition cost

Blended  $31,000 ÷ $618 = 50.2
Paid only  $31,000 ÷ $1,600 = 19.4

50:1  and  19:1

Computed on the rounded revenue figure. The portfolio figure is stated as near $31,000. Both results are gross revenue multiples against acquisition cost. Neither is margin. Neither accounts for clinical delivery cost, length of stay variance, bad debt or reimbursement recovery.

Benchmark against the 19:1, not the 50:1. The paid figure is the one an operator can move this quarter, and it is the one that compares cleanly across programs regardless of how developed their nonpaid channels are. The blended figure is an outcome of years of accumulated work. Quoting it as a target to a program starting from zero is misleading.

06

How families find treatment inside AI assistants

A growing share of treatment research now happens inside an AI assistant rather than on a results page. A family describes a situation, names a city, and receives a synthesized answer naming a handful of facilities with sources attached.

Appearing in that answer is a distinct outcome from ranking in search results, and it is not produced by the same work. A page engineered to rank for a commercial query is frequently not the page an assistant will cite, because assistants cite pages that answer the question asked and carry claims that can be attributed.

Almost nobody in this category measures it. The measurement is not difficult.

Prompt set
25 fixed prompts, written the way a family member or a referring clinician would actually phrase the question, held constant between runs so results stay comparable over time
Engines
5 assistants, queried without personalization or prior conversation context
Observations
125 prompt and engine pairs per measurement run
Scoring
One binary per observation. Does a specific named facility appear as a cited or linked source in the response? Mentions without attribution are not counted

Result

Citation rate, measured portfolio

48 citations ÷ 125 observations = 0.384

38.4%

A visibility measurement, not an attributed admission figure. AI assistant referrals are not separately attributable in most client analytics stacks today. They land inside direct and organic traffic. The methodology covers how that is being fixed for the next edition.

What moves it

  • A page that answers the question. Not a service page. The queries families actually ask are situational. What to do when someone refuses treatment, what insurance covers at a given level of care, what happens in the first 72 hours.
  • Claims that can be attributed. Named, dated, sourced statements get cited. Unsourced superlatives do not, and increasingly cost the whole page its credibility with the model.
  • Third party corroboration. A claim that appears only on the facility's own domain is weaker to an assistant than one it can verify elsewhere.

The reason to act on this now is cost. This channel has no per admission media cost, the competitive field is nearly empty, and the work that earns citations is the same work that earns organic search, so it is not an additional budget line. That combination will not last.

07

What operators should do about it

  1. Restate your reporting in admissions

    Not leads, not calls, not form fills. If your agency cannot produce cost per admission, that is the finding.

  2. Measure your lead to admission rate before you change any media

    Section 01 is the reason. The spread in that worked example is structural, not invented, and a media change made before you know your own rate is a guess.

  3. Calculate your paid share of admissions

    Run the section 04 arithmetic on your own two figures. If your blended and paid costs per admission are close to identical, your paid share is near 100%, and that single fact explains your acquisition cost better than anything in your ad accounts. We will run it for you.

  4. Audit intake response time against admission rate

    Lead to admission conversion is bounded by how fast someone answers. No amount of media spend compensates for an unanswered call at 9pm, and this is the cheapest fix available to most programs.

  5. Run the AI citation measurement once this quarter

    Twenty five prompts across five assistants is an afternoon of work. It establishes a baseline you can measure against next quarter, which is more than almost any competitor currently has.

Before hiring an agency for any of this, check it against the ten criteria for choosing a rehab marketing agency.

08

Methodology

This section exists so the figures above can be checked rather than taken on trust. Every number in this report is either a measured portfolio figure defined below, or a derivation computed from those figures with the arithmetic shown inline on the page. No derived figure introduces an outside input.

There is one exception, and it is confined to section 03. The published benchmarks in section 03 are outside figures, attributed to their sources and linked. One derivation in that section divides Webserv's published figure by the paid figure in this report, so that single result does depend on an outside input, and it is labelled where it appears. Nothing in the at a glance table, and no figure in any other section, depends on anything published outside this report.

Definitions

Admission
A patient admitted to a level of care at a participating program, counted at admission date and reconciled against the program's own census. Not counted at inquiry, assessment, or insurance verification.
Media spend
Platform reported spend in managed advertising accounts, inclusive of all paid search, paid social and paid display.
Blended cost per admission
Total acquisition cost across all channels divided by all attributed admissions in the period.
Paid cost per admission
Paid media spend divided by admissions attributed to paid channels in the period.
Revenue per admission
Gross charges per admission, averaged across the portfolio. Not collected revenue.
Partner retention
Share of partner programs active at the start of a year still active at its end, reported as a rate.

Attribution

Paid admissions are attributed on platform reported conversion and reconciled against program census on a monthly cycle. Nonpaid admissions are attributed on the last source that was not direct, measured at the inquiry event. AI assistant referrals are not separately attributable in most client analytics stacks and are presently counted within direct and organic. The citation rate in section 06 is therefore a visibility measurement and is deliberately not presented as an admission source.

Data handling

Advertising, call tracking and admissions data are unified and processed inside Aellē, which operates under HIPAA, 42 CFR Part 2 and SOC 2 Type 2 compliant controls. No patient level information is used in this report or derivable from it. Every figure is aggregated across participating programs.

Period discipline

Portfolio totals, meaning $64M+ media and 37,200+ admissions, are cumulative across 2018 to the current period. Unit economics, meaning $618, $1,600 and near $31,000, are current reporting period. The two are not interchangeable. Multiplying a cumulative total by a current period rate produces a figure this report does not support, and the difference between $1,720 lifetime and $618 current in section 02 is exactly why.

Not reported in this edition

Withheld pending instrumentation

  • Channel breakdown beyond the paid and nonpaid split
  • Level of care distribution and per level acquisition economics
  • Program level ranges and dispersion
  • Payer mix
  • Cost variance across the 27 markets

Each of these requires measurement that is being built rather than estimated. Publishing a modeled figure in place of a measured one would make this report less useful every year instead of more, which is the opposite of its purpose.

Limitations

  • Revenue per admission is a portfolio average and is not representative of any single program. Gross charges differ substantially across levels of care.
  • The implied paid share in section 04 is an upper bound, for the reason stated there.
  • Partner retention is reported as a rate, not a cohort survival curve.
  • The portfolio is weighted toward residential and outpatient programs in the states covered. It is not a national sample and should not be read as one.
  • The AI citation rate reflects one measurement run against one prompt set. Assistant behavior changes between model releases, which is why the prompt set is frozen.

What the 2027 edition adds

  • Cost per admission by individual channel, with AI assistant referrals separately attributed through a dedicated parameter and an intake confirmed source field
  • Level of care distribution and per level acquisition economics
  • Lead to admission conversion benchmarks by channel
  • Intake response time measured against admission rate
  • Market level cost variance across the full footprint
  • Year over year comparison against this edition, which is why the definitions above are stated in full and will not be changed

Download the full report (PDF)

09

Questions this report answers

What is a good cost per admission for a behavioral health program?

Across the measured portfolio, blended cost per admission is $618 and paid media only cost per admission is $1,600. An operator buying admissions exclusively through paid media should benchmark against $1,600. One with developed organic, referral and AI citation channels should benchmark against the blended figure. Comparing a paid only program to a blended benchmark will always look like failure, and comparing the reverse will always look like success. Other published figures in this category vary widely and rest on different definitions. Active Marketing cites an industry range of $1,650 to $12,000 per patient on total marketing spend. Webserv reports $16,608 per admit from $16.5M of PPC spend across 994 attributed admissions in 2025, which is the same construction as the $1,600 here. Lead to Recovery publishes $600 to $914 per admission on organic search at individual client centers, a single channel result rather than a portfolio figure. Network status, payer mix and level of care all move the figure, so compare definitions before comparing numbers.

How does this compare to what other rehab marketing agencies report?

Webserv reports $16,608 per admit, built from $16.5M of PPC spend and 994 attributed admissions across its full client base in 2025. That is the same construction as the $1,600 paid figure in this report, which makes it a like for like comparison, and it works out to 10.4 times. Lead to Recovery publishes per campaign figures of $600 to $914 per admission on organic search at individual client centers, which is a single channel result at one program rather than a portfolio figure. Active Marketing cites an industry range of $1,650 to $12,000 per patient on total marketing spend, a broader measure that is not directly comparable to either. Attribution method sits underneath every one of these figures and is not published to the same depth by every source, so the construction is comparable even where the rigour cannot be verified.

Why is cost per lead a bad metric for treatment center marketing?

Because lead to admission conversion is a property of the program, not the media, and it varies by an order of magnitude between facilities. In a worked example with hypothetical inputs, two programs paying an identical $200 per lead would pay $1,667 and $10,000 per admission at 12% and 2% conversion respectively. The media is identical. The economics are sixfold apart. Cost per lead cannot distinguish between them, which makes it useless for comparison and easy to report well while performance declines.

What percentage of treatment center admissions come from paid advertising?

In this portfolio, no more than 39%. That figure is forced by arithmetic rather than surveyed. If blended cost per admission is $618 and paid cost per admission is $1,600, and nonpaid channels carry no per admission media cost, paid media can account for at most 618 divided by 1,600, or 38.6% of admissions. Allocating fixed costs lowers it further. The remaining 61% or more arrives through organic search, AI citation, referral relationships, direct demand and returning alumni.

What is a typical return on ad spend for a treatment center?

At revenue near $31,000 per admission, this portfolio returns approximately 19:1 on paid media and 50:1 blended across all channels. Both are gross revenue multiples against acquisition cost, not margin, and neither accounts for clinical delivery cost, length of stay variance, bad debt or reimbursement recovery. Operators should benchmark against the 19:1 paid figure, because it compares cleanly regardless of how developed a program's nonpaid channels are.

How much does it cost to acquire a patient admission in behavioral health?

Between $618 and $1,600 in this portfolio, depending on channel mix. Over the full managed history since 2018, $64M+ in media produced 37,200+ attributed admissions, an average of approximately $1,720 per admission. The gap between that lifetime average and the current blended figure of $618 reflects eight years of building channels that carry no marginal cost per admission, not an improvement in bidding.

How do you measure whether AI assistants recommend your facility?

Run a fixed prompt set across multiple assistants and score citations. This report uses 25 prompts, written the way a family member or referring clinician would phrase the question, run across 5 assistants without personalization, producing 125 prompt and engine observations per run. Each observation is scored on one binary. Does a specific named facility appear as a cited or linked source? Mentions without attribution do not count. Holding the prompt set constant is what makes runs comparable over time.

What is a good AI citation rate for a treatment center?

The measured portfolio returns 38.4%, or 48 citations across 125 observations. There is no published industry benchmark for this metric because almost no operator in behavioral health measures it. Any operator establishing a baseline this quarter has more measurement of this channel than nearly all of their competitors.

How is an admission attributed to a marketing channel in this report?

Paid admissions are attributed on platform reported conversion and reconciled against program census monthly. Nonpaid admissions are attributed on the last source that was not direct, measured at the inquiry event. Admissions are counted at admission date against the program's own census, never at inquiry, assessment or insurance verification. AI assistant referrals are not separately attributable in most client analytics stacks today and are counted inside direct and organic, which is why the 38.4% citation rate is presented as a visibility measurement and not as an admission source.

Does this report contain patient data?

No. Advertising, call tracking and admissions data are unified and processed inside Aellē, which operates under HIPAA, 42 CFR Part 2 and SOC 2 Type 2 compliant controls. No patient level information is used in this report or derivable from it. Every figure is aggregated across participating programs.

How often is this report published?

Annually. This is the first edition, published 14 November 2026. The definitions in section 08 are fixed so that each edition can be compared directly against the last. The 2027 edition adds per channel cost per admission, level of care economics, lead to admission conversion benchmarks by channel, intake response time against admission rate, and market level cost variance.

How to cite this report

Aellē Digital. The 2026 Behavioral Health Patient Acquisition Report, first annual edition. Published 14 November 2026. https://aelle.io/research/2026-behavioral-health-patient-acquisition-report/

Figures may be reproduced with attribution to Aellē Digital and a link to this page. Derived figures should be cited with the arithmetic shown in the section they appear in, so a reader can verify them independently.

Benchmark

Run these numbers against your own

Send your figures and we return your implied paid share, where you sit against $618, $1,600 and 19:1, and which of the five actions applies to you. No obligation and no deck.

If you do not know your blended and paid cost per admission, leave them empty. That is itself the finding, and it is the most common answer we get.

Send business figures only. Do not include any patient information in this form.

Aellē Digital. Patient acquisition infrastructure for behavioral health. Century City, Los Angeles.

Published 14 November 2026. First annual edition. Figures are current as of the stated reporting period and are drawn from managed Google and Meta advertising accounts and participating program admissions reporting, 2018 through present.

Definitions in section 08 are fixed for the purpose of year over year comparison. Corrections and methodology questions are welcome and will be reflected in the next edition.