5
Options compared
0.39
Adopted: value per $1 at risk
0.95
Best: out-licensing
4
Disbenefits recorded
Contents
1. Five Options, Compared
At Gate 0 five courses were genuinely available. The business case argues for the one that was chosen; this compares it against the four that were not.
| Option | Cost to Vitalis | Risk-adjusted value | Value per $1 at risk | Retained upside | Risk carried | Assessment |
|---|---|---|---|---|---|---|
| Kill at Gate 0 | $0 | $0 | — | None | None | ⚠ The baseline every other option is measured against, and the one most often left out. Zero cost, zero return, and the capability built over 3.7 years of discovery is written off with it. |
| Develop internally (adopted) | $243,040,000 | $96,000,000 | 0.39 | 100% | All of it | ⚠⚠ Expected value BELOW the authorized cost. Justified as buying the OPTION to continue, not the asset — and only defensible because the money is released one stage at a time. |
| Out-license after Phase 2 | $74,400,000 | $71,000,000 | 0.95 | ~15% royalty | Partner carries Phase 3 | Spend through Gate 3 only, then license. ⚠ Removes the largest risk and most of the return. The economics are close to internal development on a risk-adjusted basis and the distributions are completely different. |
| Co-develop with a partner | $121,500,000 | $84,000,000 | 0.69 | ~50% | Shared | Halves cost and upside. ⚠ Adds a governance layer with joint decision rights — the hidden cost is that gate decisions become negotiations. |
| Sell the asset at Gate 3 | $74,400,000 | $62,000,000 | 0.83 | None | Transferred entirely | A clean exit with a known number. ⚠ Forecloses everything, and a company that sells every asset at Phase 2 has no pipeline and eventually no reason to run discovery. |
The adopted option has the highest absolute value and the worst return per dollar at
risk.
Internal development returns 0.39 of risk-adjusted value for every dollar committed. Out-licensing after Phase 2 returns 0.95 — roughly 70% of the value for 30% of the capital.
That trade is the whole point of running the comparison, and it is asserted in the fact base: if the adopted option were also the most capital-efficient, it would win on every measure and the analysis would have taught nobody anything.
Internal development returns 0.39 of risk-adjusted value for every dollar committed. Out-licensing after Phase 2 returns 0.95 — roughly 70% of the value for 30% of the capital.
That trade is the whole point of running the comparison, and it is asserted in the fact base: if the adopted option were also the most capital-efficient, it would win on every measure and the analysis would have taught nobody anything.
The choice made was to accept the worse capital efficiency in exchange for 100% of the upside and complete control of the asset. That is a defensible strategic position for a company that wants to own a marketed product, and it is the wrong answer for one optimizing return on capital. The CBA does not resolve that; it makes the trade explicit so the Committee decides it deliberately.
2. The Two Options Nobody Wants to Write Down
| Why it belongs in the analysis anyway | |
|---|---|
| Kill at Gate 0 | ⚠ The baseline every other option is measured against. Without it, the comparison has no zero and every option looks positive. It is also the option a governance system exists to keep available — KILL is not failure, and a portfolio where nothing is ever killed at Gate 0 is not selecting, it is queueing. |
| Sell the asset at Gate 3 | A clean exit at $62,000,000 risk-adjusted, with every downstream risk transferred. ⚠ It forecloses everything — and a company that sells every asset after Phase 2 has no marketed products, eventually no reason to run discovery, and a valuation built entirely on other people's execution. |
Both are usually absent from a real cost-benefit analysis, and their absence is what turns one
into a business case with more tables.
An analysis that compares three ways of proceeding has already decided to proceed. The uncomfortable options are the ones that make it a comparison — and writing them down costs nothing at Gate 0, when nobody is invested yet.
By Gate 4 the same list would be politically impossible to table. The moment to write down the option of stopping is the moment stopping is still ordinary.
An analysis that compares three ways of proceeding has already decided to proceed. The uncomfortable options are the ones that make it a comparison — and writing them down costs nothing at Gate 0, when nobody is invested yet.
By Gate 4 the same list would be politically impossible to table. The moment to write down the option of stopping is the moment stopping is still ordinary.
3. Benefits
| Benefit | Type | When it lands | Note |
|---|---|---|---|
| Financial return on approval | Financial | 2031 onward | The success-case NPV. ⚠ Conditional, and the condition does most of the work. |
| A registered NME and an approved US label | Strategic | At approval | An asset the company owns outright, with 5 years NCE exclusivity and patent life to 2039. |
| First in-house 505(b)(1) capability | Institutional | Built through Stages 1–5 | ⚠ Not in the Gate 0 case. The company can now run a registrational program without a partner — and it is the benefit most likely to outlast the product. |
| A functioning stage-gate governance system | Institutional | Built through the program | ⚠ Also not in the Gate 0 case. Reusable across the pipeline. |
| Patients treated | Societal | From launch | Real and deliberately not monetized here. Assigning a dollar value to it would make the case look better and the analysis worse. |
Two of the five benefits were never in the Gate 0 case, and they are the two most likely to
outlast the product.
A first in-house 505(b)(1) capability and a functioning stage-gate governance system are both institutional — they belong to the company rather than to the molecule, and they survive whatever happens commercially. Neither was forecast, because a business case written to justify a molecule does not think to count what building it teaches the organization.
The patients-treated benefit is deliberately not monetized. A dollar figure could be constructed and it would make the case look stronger and the analysis weaker — every assumption in it would be unfalsifiable, and it would sit alongside numbers that are not.
A first in-house 505(b)(1) capability and a functioning stage-gate governance system are both institutional — they belong to the company rather than to the molecule, and they survive whatever happens commercially. Neither was forecast, because a business case written to justify a molecule does not think to count what building it teaches the organization.
The patients-treated benefit is deliberately not monetized. A dollar figure could be constructed and it would make the case look stronger and the analysis weaker — every assumption in it would be unfalsifiable, and it would sit alongside numbers that are not.
4. Disbenefits
Negative, foreseeable, and accepted anyway. Not risks — a risk might not happen. Not issues — nobody intends to fix these.
| Disbenefit | Foreseeable at Gate 0? | Assessment |
|---|---|---|
| Opportunity cost of the R&D budget | Foreseeable at Gate 0 | $243,040,000 not spent on other pipeline candidates over 7.6 years. ⚠ Recorded because a cost-benefit analysis that omits the alternative use of the money is not one. |
| Single-source manufacturing dependency | Foreseeable at Gate 0 | Accepted deliberately; a second site does not fit the ceiling. |
| Organizational attention | Foreseeable, not recorded at Gate 0 | ⚠ 109 people and the Committee's agenda for seven years. The company can run one program of this size at a time, and that constraint appears in no register. |
| Fourth-to-market commercial position | ⚠ Foreseeable at Gate 0 and NOT recorded | ⚠⚠ The disbenefit that mattered. It was knowable at candidate selection, it was not written down, and it is the one the outcome turned on. |
The outcome turned on the fourth of those, and it was knowable at candidate selection.
Being fourth into a well-characterized class was a fact in 2022. It appears in the competitive assessment, it shaped the target product profile, and it was never recorded as a disbenefit — because the Gate 0 template had no disbenefit section, and a template with no disbenefit section never grows one.
A governance system is very good at scrutinizing the things it has a register for. The section has to exist at authorization, when there is nothing unwelcome to put in it yet, because that is the only time anyone will agree to add it.
Being fourth into a well-characterized class was a fact in 2022. It appears in the competitive assessment, it shaped the target product profile, and it was never recorded as a disbenefit — because the Gate 0 template had no disbenefit section, and a template with no disbenefit section never grows one.
A governance system is very good at scrutinizing the things it has a register for. The section has to exist at authorization, when there is nothing unwelcome to put in it yet, because that is the only time anyone will agree to add it.
The third row is the quieter one. 109 people and the Committee's agenda for seven years is a real constraint — the company can run one program of this size at a time — and it appears in no register anywhere in the suite. It is recorded here because a cost-benefit analysis that counts only the money has counted the easier half.
5. How This Reads at Each Gate
The same analysis, re-read at four points, gives four different answers — and only the first one was actionable.
| Read at | What the comparison says | Still actionable? |
|---|---|---|
| Gate 0 | All five options open. Internal development is the highest-value and least capital-efficient choice. | ⚠ Yes. This is the only moment the full comparison means anything. |
| Gate 3 | Out-licensing and selling are still live; the sunk $74,400,000 is irrelevant to the forward decision. | Yes, and this is the decision point most programs skip. The question was not asked. |
| Gate 4 | ⚠ Partnering is theoretically possible and practically not — a partner joining at Phase 3 initiation buys the expensive half. | Barely. The economics have moved against every alternative. |
| Gate 5 | One option remains: submit. | No. The comparison is now a historical document. |
Gate 3 is the decision point this program did not use.
At Gate 3 the Phase 2 result was in hand, $74,400,000 was spent and unrecoverable, and the forward question was whether to commit $125,800,000 or license the asset to somebody who would. That is exactly the comparison this document exists for, and it was not re-run — the gate assessed readiness to proceed rather than whether proceeding was the best use of the next tranche.
Neither is wrong, and they are different questions. A gate that only ever asks “are we ready?” will never ask “should somebody else be doing this?”
At Gate 3 the Phase 2 result was in hand, $74,400,000 was spent and unrecoverable, and the forward question was whether to commit $125,800,000 or license the asset to somebody who would. That is exactly the comparison this document exists for, and it was not re-run — the gate assessed readiness to proceed rather than whether proceeding was the best use of the next tranche.
Neither is wrong, and they are different questions. A gate that only ever asks “are we ready?” will never ask “should somebody else be doing this?”
The transferable practice is small: re-run the option comparison at every gate that releases a material tranche, with sunk cost excluded and the alternatives re-priced. It takes an hour, it will almost always confirm the current course, and the one time it does not is worth more than every hour spent on the ones that did.
6. What Would Have Changed the Answer
Sensitivity on the decision rather than on the number. Under which conditions does a different option win?
| If this had been different | The answer becomes | Why |
|---|---|---|
| Probability of success below ~5% | Out-license after Phase 2 | ⚠ At 7.0% the expected value is already below the authorized cost. Two points lower and the capital at risk stops being defensible on any reading. |
| Second or third to market rather than fourth | Internal development, decisively | ⚠⚠ The share assumption is the model's largest driver. An earlier entrant needs no differentiating claim to hold formulary position, and the whole access problem disappears. |
| A partner offering more than ~60% of economics | Co-develop | At 50% the governance cost is not worth it — gate decisions become negotiations. Above 60% the risk transfer starts to pay for the friction. |
| A second asset competing for the same budget | ⚠ Any option that frees capital | The opportunity cost is recorded as a disbenefit and priced at zero, because there was no competing candidate. With one, out-licensing wins on capital efficiency alone. |
| A head-to-head trial affordable inside the ceiling | Internal, and comfortably | It would have retired the differentiation risk that the whole comparison hinges on. ⚠ It was priced at $34,000,000 and declined — see CR-D1. |
Two of the five conditions were already known at Gate 0, and neither moved the decision.
The program was fourth to market and the probability of success was 7.0% — both facts, both in the candidate assessment, and both pointing at out-licensing on a pure capital-efficiency reading. The decision to develop internally was taken with those in view, for a strategic reason that is legitimate and was never written down as such: the company wanted to own a marketed product and build the capability to register one.
***A strategic rationale that is real but unrecorded looks, five years later, exactly like an analysis nobody did.*** The institutional benefits in §3 are that rationale, recovered after the fact — and they would have been worth far more stated at Gate 0, where they would have had to survive being examined.
The program was fourth to market and the probability of success was 7.0% — both facts, both in the candidate assessment, and both pointing at out-licensing on a pure capital-efficiency reading. The decision to develop internally was taken with those in view, for a strategic reason that is legitimate and was never written down as such: the company wanted to own a marketed product and build the capability to register one.
***A strategic rationale that is real but unrecorded looks, five years later, exactly like an analysis nobody did.*** The institutional benefits in §3 are that rationale, recovered after the fact — and they would have been worth far more stated at Gate 0, where they would have had to survive being examined.