Use case · portfolio triage
Rank the pipeline on the same evidence.
When five programs compete for one budget, the comparison is usually between five differently-built cases. Run all of them through the same six pillars, from the same public sources, and the ranking becomes about the assets rather than about the decks.
The same six pillars, applied to every asset.
The situation
Five programs, five incomparable cases.
Each program is presented by the team that owns it, with its own analysis, its own assumptions and its own definition of a comparable trial. The portfolio decision is then made across cases that cannot actually be compared.
- Identical pillars per asset
- 6
- Analyses run the same way each time
- 30+
- Comparable outputs across the portfolio
- One scale
What happens today
The comparison is between presentations, not assets.
Portfolio review is the highest-leverage meeting in the organisation and usually the one with the least methodological discipline behind its inputs.
Each case is built differently
Different assumptions, different comparator sets, different vintages of market data. The numbers look comparable on a slide and are not.
Depth is rationed
Only the two or three programs already believed in get a deep analysis. The others are ranked on a summary, which reinforces the belief that put them there.
Ownership colours the analysis
Nobody presents the case for stopping their own program. That is human and it is also a systematic bias in every input the committee receives.
Nothing is re-derivable next cycle
At the next review the analysis is rebuilt from scratch, so a change in ranking cannot be separated from a change in method.
What you do instead
One method, applied to everything.
Every asset, the same assessment
Six pillars and more than 30 analyses run identically per asset, so the outputs sit on one scale rather than on five.
Comparable directional probabilities
Each program’s probability of success is derived the same way — the same base-rate treatment, the same adjustment logic, the same disclosure of assumptions.
Valuation on comparable inputs
Market size, value-based price and risk-adjusted eNPV computed with a consistent method, so relative value means something.
Execution risk in the ranking
Enrolment reach, eligibility burden and competing-trial density enter the comparison, rather than sitting outside it as an operational detail.
Reproducible across cycles
Runs are reproducible against their evidence snapshot, so next quarter’s change in ranking reflects a change in evidence rather than a change in analyst.
A defensible cut line
When the portfolio has to shrink, the programs below the line have a written, cited reason — which is what makes the decision survive the follow-up.
What runs, and who drafts
The scope for this job
Triage needs comparability more than depth on any single pillar. Where & How runs per asset and matters most for the programs near the cut line, where execution risk decides it.
Pillars involved
How it runs
Submit the portfolio, read one ranking.
Submit every asset
Protocols, synopses, or just molecule-and-indication pairs. Assets at different stages of definition can still be assessed on the same pillars.
The same assessment runs on each
Identical analyses, identical sources, identical treatment of assumptions — differences in the output reflect differences in the assets.
Rank, cut, and record
Compare on one scale, decide where the line falls, and sign the resulting decisions so the reasoning survives to the next cycle.
What lands on your desk
A ranking with a reason per row.
Not a score with no derivation — a ranking where each position can be opened and read down to its sources.
What lands on your desk
Portfolio comparison + per-asset assessments
- A single comparison view across assets on the six pillars.
- A directional probability of success per asset, derived identically for each.
- Risk-adjusted valuation inputs computed on a consistent method across the portfolio.
- Execution-risk indicators — enrolment reach, eligibility burden, competing-trial density — per asset.
- The full cited assessment behind every row, openable from the ranking.
- Signed decision records for the programs where a call was made.
Honest limits
Where this stops
What this does not do for you. Naming it here is cheaper for both of us than finding it in a procurement review.
It compares assets on public evidence. Strategic fit, internal capability, existing commitments and portfolio balance are yours to weigh alongside it.
Probabilities are directional and are comparable to each other because the method is identical — that is not the same as being calibrated in absolute terms.
Assets at very different stages produce assessments of different depth. The comparison view says where the evidence is thin rather than normalising the gap away.
It does not know your internal data, so a program whose case rests on unpublished results will look thinner here than it is.
A ranking is an input to a portfolio decision, not the decision.
Questions
The ones you would ask first
How many assets can we run?
Portfolio screening is a supported mode rather than a batch of one-off runs — assets are submitted together and returned on one comparison view.
Our assets are at different stages. Does the comparison still hold?
The pillars are the same for each, and the assessment states where evidence is thin for an earlier asset instead of quietly filling the gap. That gap is itself decision-relevant.
Are the probabilities comparable across assets?
They are derived identically, which is what makes relative comparison meaningful. They remain directional — comparability across the portfolio is the claim, absolute calibration is not.
Can we re-run the triage next quarter and see what moved?
Yes. Runs are reproducible against their evidence snapshot, so a change in position can be attributed to a change in the evidence rather than to a change in method.
Compare the pipeline on one scale.
Send us the portfolio — protocols, synopses, or just molecules and indications. We’ll run every asset through the same six pillars and return one cited comparison.