Fractional value bidding assigns partial conversion values to meaningful micro-actions — qualified lead, demo booked, high-intent page view — rather than counting only completed conversions. It lets bidding optimize on the full value of a journey, especially where conversions are rare or delayed, so the algorithm learns from signals it would otherwise ignore between zero and a full conversion.
Most bidding treats conversion as binary: either it happened and is worth its full value, or it didn’t and is worth nothing. But journeys aren’t binary. A visitor who qualified, booked a demo, or spent ten minutes on your pricing page did something genuinely valuable — they just haven’t crossed the finish line yet. Standard bidding sees all of that as zero, which means the algorithm learns nothing from the most informative signals in the funnel. For more on query control, see our approach to negative keywords.
Fractional value bidding fixes this by assigning partial worth to those in-between actions, so bidding optimizes on the full shape of the journey — not just the rare moment of completion. It’s especially powerful where full conversions are sparse or take weeks to arrive.
All-or-nothing vs. fractional
The difference is whether the algorithm can see value in the journey or only at its endpoint. To understand the underlying data infrastructure, review our guide on server-side tagging.
| Binary | Fractional | |
|---|---|---|
| Values | Full conversions only | Journey signals too |
| Micro-actions worth | Zero | Partial value |
| Learns from | Rare endpoints | Whole funnel |
| Best when | Conversions frequent | Conversions rare / slow |
Why partial value helps the algorithm
Smart bidding needs signal to learn, and when full conversions are rare or delayed, it’s effectively starved — optimizing on a trickle of endpoints. Fractional values give it a richer, denser signal: every qualified lead or high-intent action becomes a weighted data point, so the algorithm can find patterns long before a full conversion would have taught it anything. More signal, sooner, means faster and better optimization.
Micro-actions and their fractional weight
Illustrative partial values toward a full conversion.
How to assign fractional values
The work is calibration: identify the meaningful steps in your journey, then assign each a partial value proportional to how often it leads to a full conversion and what that conversion is worth. A demo that closes 40% of the time at $10k carries real fractional value; a newsletter signup carries a little. Feed those weighted values to bidding and the algorithm optimizes toward the actions that actually predict revenue — not just the final sale.
Isn’t this just micro-conversion tracking?
Tracking micro-conversions counts them; fractional value bidding weights them by their real contribution and feeds those weights into bidding. The discipline is in the calibration — a micro-action valued wrongly misleads the algorithm just like any bad signal. Weighting, not just counting, is the point.
When full conversions are rare or slow, bidding on them alone leaves the algorithm half-blind. Fractional value bidding hands it the whole journey, properly weighted — so it learns from every meaningful step toward the sale, not just the rare moment it arrives.
This article is a spoke node connected to our core technical hubs. To explore the broader architecture, visit our primary pillar pages: