Paid Advertising

Platform-reported ROAS can overcount as Meta attributes on a 7-day click window and takes credit for conversions that would have happened anyway. In a healthcare environment it's worse, because data restrictions mean fewer signals flow back to the platform in the first place.

The fix is to build your own attribution layer. Connect UTM parameters directly to your CRM so every booking, consultation, or purchase can be traced back to the campaign and creative that drove it. Use platform data as a directional signal for in-platform optimisation decisions, GA4 for a cross-channel view, and CRM revenue as the actual source of truth for budget decisions.

At Ouronyx, I built a UTM-to-CRM bridge into Cliniko. It revealed campaigns that looked efficient on Meta but were generating low-value enquiries, and others that looked average on paper but were driving the highest-AOV clients. Budget decisions moved from reported ROAS to real revenue which is how the account got to 3.64× verified against actual clinic income.

The learning phase requires roughly 50 optimisation events per ad set per week. If your account is fragmented across too many narrow ad sets, none of them accumulate enough signal to exit. The fix isn't to wait longer or lower the bid; it's to consolidate.

Reduce the number of active ad sets and broaden audiences so data density concentrates rather than spreads. Under Meta's Andromeda framework, the delivery system responds to creative signal quality so the qualification work that used to happen at audience level now happens through creative. A tighter, more specific message reaches the right person without manual audience restriction.

At Ouronyx, inherited campaigns were stuck in this exact pattern - too narrow, high CPMs, stagnant performance. Consolidating the structure and letting creative carry the targeting shifted the account out of learning, stabilised CPAs, and gave the algorithm enough signal to optimise properly.

Most paid media creative relies on a hook that's either promotional (offer, price, urgency) or comparative (before/after, results). In regulated healthcare, both are largely off the table. That forces a more rigorous process, which actually produces better creative.

Start with audience insight rather than creative production. Talk to the people closest to the customer in a clinic context that's the doctors and consultants and layer that with Reddit research, review mining, and competitor ad library analysis. The goal is to find the real fear or desire underneath the surface-level want.

At Ouronyx, every research source pointed to the same insight: women didn't want to look like they'd had work done. They wanted reassurance it would look natural. That became the entire creative direction - precision, trust, clinical confidence. "Restore and define facial features." No promotions, no noise. I filmed the first round myself in the clinic to prove the direction before spending on production. That creative ran for over 12 months and became the highest impression volume asset in the account - which is the most honest measure of whether a creative direction works.