What to Look For in Facebook Ad Management Services

The Evaluation Has Moved Past Creative and CPM

The vendors worth interviewing in 2026 are not the ones showing thumbnails of high-CTR video ads. They are the ones who can walk a treatment center CMO through a pixel data map, a substantiation file for every objective claim in creative, a call-recording consent workflow, and a qualified-call attribution model that reconciles with the admissions team’s CRM. Facebook ad management for addiction treatment and behavioral health has become a compliance-and-measurement discipline first, and a creative-buying discipline second.

The last three years have made that reordering explicit. The FTC’s April 2024 action against Monument banned an alcohol-treatment provider from disclosing consumer health data to ad platforms including Meta and Google after alleging that disclosure occurred without appropriate consent ¹³. In June 2025, Evoke Wellness agreed to pay $1.9 million to settle FTC claims involving misleading SUD treatment representations, competitor-name search bidding, and call-center conduct — a settlement that reached well past the ad unit itself ¹⁴. Earlier, the FTC challenged a clinic’s “98% Improvement Rate” and “rapid, painless, effective, and safe” language for lack of competent and reliable scientific support ¹⁵.

A CMO evaluating a Facebook ad management partner should score them against that record. The sections below define the specific capabilities — creative substantiation, tracking and consent architecture, post-click accountability, and qualified-call measurement — that separate a defensible operator from a lead-volume shop.

Creative Substantiation: The Standard Regulators Actually Apply

Pixel, Consent, and the Data Architecture Behind the Ad

The Meta Pixel is where most treatment-center Facebook programs generate their largest hidden liability. In July 2023, the FTC and HHS jointly warned approximately 130 hospital systems and telehealth providers that tracking technologies on their sites and apps may impermissibly disclose sensitive health information to third parties, including advertising vendors ⁵. That warning is the baseline scope of regulator attention a Facebook ad management vendor is now expected to understand before touching a pixel install.

A competent vendor performs a technical audit before optimization work begins. That audit inventories every tracker on public pages, appointment portals, chat widgets, and authenticated environments; documents which events fire on which URLs; and inspects the parameter payload — form field values, URL query strings, referrer data, and custom event names — that leaves the browser for Meta’s servers ⁴. HHS treats tracking on authenticated pages as regulated by the Privacy and Security Rules, and does not assume that an unauthenticated treatment page is automatically safe; the fact pattern depends on what the pixel actually collects ⁴.

The FTC’s tracking-pixel analysis raises the operational bar further. Impermissible pixel disclosures can implicate the FTC Act, the Health Breach Notification Rule, HIPAA, state privacy statutes, and a company’s own privacy promises ⁶. Monument’s April 2024 settlement made that risk concrete for the addiction-treatment category specifically: the FTC alleged that the alcohol-treatment provider disclosed users’ personal health data to Meta and Google for advertising without appropriate consent, and the proposed order banned such disclosures going forward ¹³.

Ask each shortlisted vendor for four specific artifacts:

1. A pixel-and-event map showing every domain, page template, and event parameter transmitted to Meta.
2. A written consent architecture describing how affirmative choice is captured before non-essential tracking fires, and how that choice is recorded and suppressed downstream.
3. Executed vendor contracts and data-processing terms with any measurement partner, including retention limits and permitted uses.
4. A proposal for privacy-preserving measurement — server-side Conversions API with parameter filtering, hashed identifiers, and offline conversion uploads sourced from consented CRM data — rather than a browser-side pixel firing on high-intent pages ^{4, 6}.

The Post-Click Accountability Chain

Measurement: Qualified Calls Beat Lead Volume

Lead volume inside Ads Manager is the wrong scoreboard for a treatment center. The metric that pays payroll is a qualified inquiry that reaches admissions, verifies benefits, and shows up for assessment — and that number rarely tracks the on-platform lead count in a straight line.

A family drug-use prevention recruitment study makes the volume-versus-quality gap concrete. Facebook ads produced 70.4% of screened participants and 65.1% of eligible participants, yet eligibility and consent rates were lower among Facebook-recruited participants than among those who came in through referral sources ¹⁰. The study measured research recruitment for a prevention program, not treatment admissions, so the numbers should not be read as a CPL benchmark. The pattern is the transferable finding: a channel can dominate top-of-funnel share while producing a downstream population that qualifies at a lower rate than other sources.

A competent vendor builds reporting around that reality. Meta’s in-platform lead count is one column. The columns that matter sit further down:

• Connected calls above a defined talk-time threshold
• VOB-eligible inquiries
• Clinical-fit qualifications from the admissions team
• Scheduled assessments
• Admissions with source attribution reconciled against the CRM

Each stage should be reported with a conversion rate against the prior stage, not against impressions.

Attribution has to be built, not assumed. Ask each shortlisted vendor how it handles call tracking with dynamic number insertion, how it suppresses duplicate leads across form submissions and phone calls from the same household, how it reconciles offline admissions data back to campaign and ad-set level through the Conversions API, and how it treats view-through credit versus click-based credit. A vendor that cannot describe those mechanics in writing is optimizing toward the metric Meta rewards, not the metric the CFO tracks.

Set the reporting cadence around admissions math. A weekly dashboard should show cost per connected call, cost per qualified inquiry, and cost per admission by campaign, with the qualification rate at each handoff surfaced next to spend. When those rates move, the creative, audience, and landing-page changes that preceded the shift should be documented in the same report. That is what accountability looks like when the objective is census, not lead forms.

Platform Performance Varies by Audience and Objective

Any vendor pitching a fixed cost-per-lead across Meta placements is describing a media buy that does not exist. A 2025 peer-reviewed comparison of Facebook ads and Instagram promoted posts recruiting 1,216 sexual-minority adolescents into a drug-prevention trial illustrates why. The campaign spent $25,400.31 at $20.89 per participant across both platforms. Facebook ads generated 51% fewer clicks, a 338% higher cost per click, and a 90% lower click-through rate than Instagram promoted posts targeting the same audience with the same objective ⁸.

That delta was produced inside a single research study with one audience, one creative approach, and one recruitment objective — not a treatment-admissions campaign. The numbers should not be handed to an admissions team as a benchmark for detox or residential inquiries. The transferable finding is directional: the same dollar routed to two Meta placements produced order-of-magnitude differences in click economics, and the winning placement was not the one most treatment brands default to.

A competent vendor turns that variance into a testing protocol rather than a pitch-deck promise. Expect written test plans that hold audience, objective, and offer constant while varying one lever at a time:

• Placement (Feed, Reels, Stories, Audience Network)
• Creative format (static, short video, carousel)
• Message frame

Expect statistical stopping rules, minimum spend thresholds per cell, and a defined qualification metric downstream of the click — connected calls or VOB-eligible inquiries — before a winner is declared.

Push back on any proposal that quotes a cost per lead without naming the placement, audience definition, creative type, and qualification stage it was measured against. Numbers without that scope are marketing copy, not media planning.

Engagement Lift Is Not Help-Seeking Behavior

A Facebook campaign can move mental-health literacy without moving a single admissions call. A randomized controlled experiment of a Facebook depression-awareness campaign found that the ad improved participants’ mental-health literacy and willingness to access additional information, but produced no significant change in help-seeking attitudes compared with the control group. The same study reported that a positive-face advertisement generated more click-throughs than a sad-face advertisement ⁹.

That gap between knowing more, clicking more, and actually seeking treatment is the trap a CTR-focused vendor walks a treatment center into. A creative that wins the placement auction on engagement may be recruiting scrollers who are learning about depression, not prospects who are calling admissions.

A competent vendor defines the qualified outcome before the first ad runs and tests creative against that outcome, not against thumb-stops. Educational and empathy-framed creative gets tested against connected calls above a talk-time threshold and VOB-eligible inquiries, with click-through and video-completion treated as diagnostic signals rather than success criteria. When a variant lifts CTR but flattens connected-call rate, it loses. Report engagement inside the funnel, not as the funnel.

A Vendor Scorecard Your Legal and Clinical Reviewers Will Recognize

The scoring model that survives a joint review with a treatment center’s general counsel, compliance officer, and clinical director is not built from marketing categories. It is built from the four risk domains that regulators and reviewers already track: creative substantiation, tracking and consent architecture, post-click accountability, and qualified-call measurement. Score each domain pass/fail against a named artifact the vendor produces, not against a claim the vendor makes in a pitch.

On creative substantiation, the passing artifact is a written evidence file citing competent and reliable scientific support for every objective claim in the ad, with a documented reviewer sign-off before the campaign queues ^{1, 16}. Fail any vendor whose approval workflow ends at brand and legal review without a per-claim citation, or whose testimonial policy does not address typicality and material-connection disclosure across Reels, Stories, and in-feed video ^{3, 15}.

On tracking and consent architecture, the passing artifacts are a pixel-and-event map covering every domain and event parameter transmitted to Meta, a written consent architecture describing how affirmative choice is captured and suppressed, executed data-processing terms with measurement partners, and a server-side measurement proposal that filters sensitive parameters ^{4, 5, 6}. The Monument order sets the disqualifying pattern: health data disclosed to ad platforms without appropriate consent ¹³.

On post-click accountability, the passing artifacts are a landing-page change-control log with substantiation citations for every hero claim, a competitor-name bidding audit, and a call-quality review process with a documented escalation path when intake specialists exceed what the ad and page support ¹⁴. For federally assisted SUD programs, add a Part 2 workflow diagram covering CRM sync, custom audience uploads, and call recordings, dated against the February 16, 2026 compliance date ¹².

If You Manage Multiple Facilities or a Portfolio

A CMO overseeing three residential facilities in different states, or a private-equity-backed portfolio spanning detox, PHP, and IOP brands, applies the same four risk domains — but the artifacts have to be produced per facility, not per contract. Meta authorization, LegitScript status, and state licensure verification live at the facility level; a shared business manager does not confer approval across brands.

Require the vendor to deliver a per-facility register:

• Pixel-and-event map
• Consent architecture
• Landing-page substantiation log
• Competitor-bidding audit
• Part 2 workflow diagram against the February 16, 2026 compliance date ¹²

Suppression lists and custom audiences should never cross facility boundaries without a documented consent basis and redisclosure analysis; a lead who called the Florida detox brand is not a lookalike seed for the California PHP ¹³.

Reporting rolls up, but qualification rates stay per facility. Cost per admission at the portfolio level hides which brand is subsidizing which, and which market is producing calls that never convert to VOB. Insist on facility-level dashboards feeding the portfolio view, not the reverse.