Smart Bidding for healthcare: tROAS vs tCPA vs Max Conversions decision tree

A clinic owner asked which bidding strategy is best. There is no best. There is a right one for your stage of data. Here is the decision tree we use to pick.

Stage 1: New account, no data (0 to 30 conversions)

Use Maximise Conversions. No target, no constraints. Just "spend the budget to get the most conversions."

This sounds dumb. It is not. Smart Bidding needs conversion volume to learn. Constraining it on a tCPA target before it has data forces it to make decisions on noise. Let it bid as it wants for 4 to 6 weeks while you accumulate the first 50 conversions.

Stage 2: 30 to 100 conversions, no value signal

Switch to Target CPA (tCPA). Set the target slightly above your current average CPA. Smart Bidding now optimises for keeping CPA at or below that number.

This works when:

• All conversions are roughly equal in value to the business.
• You want predictable cost per conversion.
• You do not have revenue data per conversion yet.

Stage 3: 100+ conversions WITH revenue values

Switch to Target ROAS (tROAS). This requires uploading conversion values, not just counts. Smart Bidding optimises for revenue, not conversion count.

tROAS works when:

• Different conversions have different value (e.g. a botox booking and an implant booking).
• You upload offline conversions from your CRM with the actual revenue.
• You can articulate a target return (e.g. "I want AED 4 of revenue for every AED 1 of ad spend").

Stage 4: Multi-product, complex business

Use Maximise Conversion Value with target ROAS. This is the most advanced version. It optimises across multiple conversion actions with different values without forcing you to think about CPA per action.

Common mistakes

Jumping straight to tROAS on a new account. No data, no learning, no result. Maximise Conversions first.

Setting tCPA too low. If your historical CPA is AED 200 and you set tCPA to AED 100, Smart Bidding restricts spend until it can hit that target. Volume crashes.

Changing strategy too often. Each switch costs 1 to 2 weeks of relearning. Switch at most once per month.

The decision tree summarised

• 0 to 30 conversions → Maximise Conversions
• 30 to 100 conversions, no values → Target CPA
• 100+ conversions with revenue values → Target ROAS
• Complex multi-product → Max Conversion Value + tROAS

The point

Smart Bidding strategy is not a permanent decision. It evolves with your data maturity. Start simple, add constraints as data accumulates, never skip ahead.

Published 5 Jun 2026 · Access One