ORTHO LAB is an orthodontic specialty clinic in Bandung with a great reputation and dentists who take the time to explain. The problem was the ads: the same three creatives had been running for almost a year, cost per chat kept climbing, and most chats stopped after asking the price. We built a weekly creative engine, taught Meta to recognize patients who actually got braces, and split the ads for teens, college students, and parents.
The resultBraces starts grew 2.3× in six months and cost per start fell 48%, with no increase in ad spend.
Meet ORTHO LAB.
drg. Bima Aditya, Sp.Ort opened ORTHO LAB seven years ago. His patients are a mix: teens brought in by their parents, college students from the campuses around Dago, and young professionals who want straighter teeth before their wedding. Referrals from existing patients have always been strong. Instagram ads were run by the clinic admin with help from a freelancer, using the same promo photos and testimonial videos since the year before.
The clinic wasn't short on chats. It was short on chats from people ready to get braces. The ads were optimized for “messages started,” so Meta looked for the people most likely to send a chat, not the ones most likely to come in for a consultation. Nobody knew which creative produced braces starts, because the clinic's appointment data never met the ad data.
The same three ads ran for 11 months. Frequency above 6, CTR down from 1.4% to 0.6%.
Teens, college students, and parents saw the same ad in the same language.
Meta learned from who sent a chat, not who got braces.
“Discount on braces” attracted bargain hunters who rarely went on to a consultation.
A creative engine that learns from patients, not likes.
New creative every week. Every creative is judged by the braces starts it produces. Losers get switched off; winners get more variations.
- 01Patient question research
We read 400 old chats and grouped the objections: does it hurt, how long does it take, installments, and fear of looking weird.
- 02Creative library by audience
Different hooks for teens (confidence), college students (installments), and parents (specialist dentist, safety).
- 036–8 creatives tested weekly
Short videos, before-and-after carousels, and creator videos. Everything is tested on a small budget first.
- 04Braces-start signals sent to Meta
Braces appointments from the clinic system are sent back through the Conversions API, so Meta learns to find people who look like real patients.
What didn’t work first.
Not everything we tried worked right away. Here are two things we stopped, and what we learned from them.
A 90-second video on types of brackets got lots of views but almost no chats. We cut it down to 15 seconds with one message.
The cheapest cost per chat, but only 1 in 19 came in for a consultation. We replaced it with clearly displayed 0% installments.
The engagement timeline.
Account audit, research on 400 chats, Conversions API set up from clinic appointments. First 12 creatives launched.
Weekly testing underway. Ads split by audience. Optimization moved from chats to scheduled consultations.
Winning creatives got more variations, and budget moved to the parent audience, which turned out to start the most braces.



