Clinical Trial Success Rates
PoS by phase × TA × modality × sponsor.
A defensible view of clinical-trial success rates — by phase, therapy area, modality and sponsor — for portfolio risk and asset valuation.
"What success rate should we apply to this asset for portfolio and valuation decisions?"
Phase 1→approval averages ~10% in pharma but varies enormously by TA and modality. Defensible PoS uses TA × phase × modality × sponsor — never a global average.
Probability of success is the single most leveraged number in pharma portfolio strategy. Using a global average is the most common mistake. Defensible PoS is always conditioned on therapy area, phase, modality and sponsor.
Build PoS as a system
Publish PoS values per asset as live numbers updated on every readout. Track the historical accuracy of your own PoS calls vs realized outcomes — that is what makes the system defensible.
What we’re seeing in the data.
Onc P1→approval is far below industry avg
Solid-tumor IO and ADC programs have unique attrition curves.
Rare disease has unusually high P3 PoS
Smaller trials, biomarker selection and accelerated pathways shift PoS up.
Big pharma sponsors lift PoS
Better trial design, regulatory experience and resource depth measurable in PoS.
Endpoint type matters
Surrogate vs hard endpoints affect both PoS and label-quality.
How to think about it.
-
01
Anchor on TA × phase baseline
Use published TA-specific PoS curves.
-
02
Adjust for modality
mAb, ADC, mRNA, ATMP, small molecule.
-
03
Adjust for sponsor track record
Big pharma vs first-time biotech.
-
04
Adjust for endpoint risk
Surrogate vs hard endpoint.
-
05
Update on each readout
PoS evolves with every interim, primary, and AdComm.
What separates a good answer from a defensible one.
Published PoS often overstates due to survivorship bias.
PoS for sub-group label may differ from overall PoS.
Some "approval" delays are manufacturing not science.
Late-stage regulatory risk doesn’t always show in trial data.
Where the signal comes from.
Common questions.
Should we use industry average PoS?
No — always use TA × phase × modality × sponsor. Industry average is misleading.
How often does PoS materially change?
On every interim or primary readout — sometimes by 30+ percentage points.
Want this answered on your data?
We build decision systems on top of analyses like this — so the next question takes minutes, not weeks.
Talk to a strategist