Most accounts optimize for one thing: more conversions, cheaper. It feels right. It’s also why so many of them plateau.
The problem is that the algorithm only knows what you tell it. If every conversion is worth “1,” it will happily find you a flood of the cheapest, lowest-intent ones it can. You hit your cost-per-conversion target and wonder why revenue didn’t move.
The fix: tell the platform what a conversion is worth
Value-based bidding replaces the binary “converted / didn’t” signal with a number that reflects actual business value. Instead of optimizing for volume, the algorithm optimizes for value, and starts spending where the valuable conversions are.
Even a rough value model beats no model:
| Action | Old signal | Value signal |
|---|---|---|
| Newsletter sign-up | 1 conversion | €5 |
| Add to cart / start booking | 1 conversion | €40 |
| Completed purchase / confirmed booking | 1 conversion | €180 |
Now the algorithm has a reason to chase the carts and bookings that actually become revenue, instead of padding the numbers with newsletter sign-ups.
Where the values come from
You don’t need a perfect LTV model on day one. Start with what you know:
- Conversion rate by action, from your analytics and back office
- Average order value or booking value
- A simple
action_value = conversion_rate × average_valueper step
Feed that back as the conversion value (server-side, so it’s accurate) and let Smart Bidding do the rest.
A real example
For a Swiss marketplace client, I mapped the full five-step booking funnel and assigned a purchase-anchored value to each step. Within weeks the campaigns shifted spend toward the segments that actually converted to bookings: same budget, better mix, more revenue. No new channels. Just a better incentive.
That’s the lesson: the algorithm isn’t broken, your signal is. Fix what you reward, and growth follows.