Rebuilding a quiz funnel to clear Meta's sensitive-health-content policy
A dental practice group's implant-suitability quiz ad kept getting paused on Meta for naming a condition or an outcome, which the platform's sensitive-health-content policy flags. The quiz was rebuilt as a multi-step form deferring the phone number to a later step, feeding a HubSpot workflow that creates the CRM record on a cleared submission. The rebuilt funnel relaunched on Meta in April 2026.
Executive Summary
Context
A multi-location dental practice group ran a quiz-style ad on Meta screening visitors for implant suitability before they reached a form. The ad named the condition and the outcome, and Meta's review paused the campaign for it each time.
What We Built
We rewrote the ad and quiz copy to drop the named condition and outcome, rebuilt the form to ask for a name and email first with the phone number deferred under an A/B split, and connected a cleared submission to a CRM-creating workflow.
Tech Stack
- HubSpot landing page and multi-step form
- HubSpot workflow creating the CRM contact record
- Meta Ads (Facebook/Instagram) quiz campaign
Not a fit unless the ad and quiz copy can describe the offer without naming the condition or outcome at all, since the gate depends on that generality rather than a case-by-case review.
The Challenge
A quiz-style ad screening patients for implant suitability ran on Meta as part of the practice group's paid-media retainer. Meta's automated review flags copy that names a personal health condition or outcome as sensitive-health content, and the original creative named both.
Each launch tripped the flag and paused delivery. So the ad sat idle until someone diagnosed the line at fault and resubmitted it, and the cycle repeated because the underlying language and form hadn't changed.
Our Approach
The original ad and quiz named the condition and the outcome directly, describing missing teeth and a restored smile, which is exactly what Meta's policy flags in a screening ad. Each relaunch of that language tripped the same review. So we kept the funnel's shape (an ad leading to a landing page, a form, a thank-you step and a CRM record) and rewrote the entry point and the form against that gate.
The copy no longer names the condition or the outcome, and it holds the offer at a generality that the review doesn't flag. The form now asks for a name and email on entry, with the phone number deferred to a later step under an A/B split.
A HubSpot workflow triggers on a completed submission and creates the CRM record. It's the same trigger the flagged version used, and it now fires on copy that clears review. The trade-off is that the offer has to stay general, because the gate depends on that rather than on a case-by-case review.
Impact
The campaign runs on Meta without repeated policy holds
The rebuilt ad and quiz no longer name the condition or the outcome, so Meta's review has nothing in the copy left to flag. The quiz relaunched in April 2026 and has kept running since, rather than sitting in review for each new variant.
A lead only needs a name and email to clear the form's first step
The phone-number field moved to a later step, tested through an A/B split against asking upfront, so a visitor only gives a name and email to get started. That lower first-step commitment was the friction the rebuild targeted.
A completed quiz creates its own CRM record
A HubSpot workflow watches for a finished submission and creates the contact record directly from it, so a completed quiz becomes a working lead in the CRM without a manual export or a separate sign-up step.
New variants go to review already clear of the flagged language
Because the rule sits in the copy brief itself, a new ad variant reaches Meta's review without the named condition or outcome, instead of waiting for someone to find the flagged line after a pause.
Ad and quiz copy were rewritten to drop the specific condition and the outcome it implies, the generality that Meta's review doesn't flag. The rule applies to the ad creative and the quiz's own questions alike.
The form asks for a first name, last name and email on entry and moves the phone number to a later step, tested through an A/B split. The later ask keeps first-step friction low while still collecting a callable number.
A HubSpot workflow triggers on a completed submission and creates the CRM contact record, the same trigger the flagged funnel used, now firing on copy that clears review.
The funnel kept its ad, landing page, form, thank-you step and CRM record structure, so the fix sat in the copy and field order rather than a new funnel built from scratch.
Meta used to pause the quiz ad on every launch, because the copy named a personal health condition and an outcome. The policy check now passes, because the copy names neither. The form asks only for a name and email, and the phone number comes later under an A/B split. A completed submission still fires the same HubSpot workflow that creates the CRM record.
FAQ
Meta's sensitive-health-content policy flags copy naming a personal health condition or outcome, and the original ad and quiz both did. Each relaunch of the same language tripped the same review, so the campaign paused again.
The rebuilt form asks for a name and email first and defers the phone number under an A/B split. A HubSpot workflow then creates the CRM record once the form clears, rather than a manual step.
