Testing plan and staged funding
Buy the information a staged entry would buy, at about a thirteenth of the price, and decide afterwards.
| Order | Test | Cost | What it settles | Evidence strength |
|---|---|---|---|---|
| 1 | Certification body gap assessment and quoted timeline | Under $0.1M | The calendar, which is what makes the 2029 bar unreachable | Strong. A quoted timeline is a commitment |
| 2 | Quotations against two module customer specifications | Under $1.0M | Module content per unit and achievable price | Strong |
| 3 | Quotations against two private-label buyer specifications | Under $0.5M | The electrode price points, and with them whether MERIDIAN's roll-to-roll cost advantage is real | Strong |
| 4 | Approaches to three private-label buyers | Under $0.5M | Whether the 20% merchant share is anywhere near right. It carries 22% of the reachable pool | Moderate |
| 5 | Cartridge volume study | Under $1.0M | The one opportunity excluded from every total here, and the best capability fit in the candidate set | Moderate |
| 6 | Shared-line utilization measurement | Under $0.5M | Whether ten roll-to-roll lines can be found without displacing the existing business | Strong |
| 7 | Approaches to three CGM challengers on second-source supply | Under $0.5M | Whether CGM merchant share is 5% or better | Moderate |
| 8 | Acquisition target screen against the four partner criteria | About $1.0M | Whether the calendar can be bought | Moderate |
The recommendation is not to stage an entry. It is to buy the information a staged entry would buy, at about a thirteenth of the price, and to decide afterwards.
A staged entry would be the obvious alternative, and it does not survive the numbers. The initiative never reaches an operating profit inside the horizon, so a first stage is not a cheap option on a marginal proposition; it is $67M spent moving toward a position that loses money. What is worth buying is narrower: four of the figures behind that answer are untested, and testing them is cheap.
The test program, about $5.1M, run before the November decision where possible and immediately after it otherwise:
Tests 1 to 3 cost under $1.6M together and settle the two figures that decide most of the answer. They are worth running whatever leadership concludes, because they also inform the existing business: a certified line and a known cost position are assets in any conversation with a medical customer, including the CGM and neurostimulation development work already in flight.
What would reopen the case, and by how much. Four results would improve it: a private-label merchant share materially above 20%; a cartridge volume that puts that segment on the scale of the injector segment; module content quotations above the modeled values; or an acquisition priced to remove the certificate from the critical path.
Their size should be stated rather than implied. Doubling the ECG merchant share adds roughly $48M of 2032 revenue at 22% gross margin. A cartridge segment the size of the injector segment adds roughly $23M at 28%. Together that is well under a tenth of the $99M gap, against a fixed organization that does not shrink. So these results would make the loss smaller; on the arithmetic here they would not close it, and none of them would make the four criteria achievable. What justifies running the tests is that they cost about a thirteenth of a first funding stage, not that they are likely to reverse the answer.
What the development work already in flight should do meanwhile. The CGM substrate and module program and the neurostimulation electrode discussion cost little, carry real technical option value, and do not require the decision this memo is about. Continue them. The recommendation here is against committing capital to a scaled position, not against the engineering.
The milestone roadmap
A milestone roadmap presupposes a decision to enter. The recommendation is to defer that decision by about six months at a cost of $5.1M, so what follows is the schedule for the evidence rather than for the entry.
- What
- Tests 1 to 3 returned
- Why
- They are cheap enough to run inside the decision window and they could change the decision
- What
- Tests 4 to 8 returned
- Why
- The remainder of the program
- What
- Reconvene on the evidence
- Why
- Either the pools are materially larger than estimated, in which case a staged entry is reconsidered against a rebuilt model, or they are not, in which case the answer stands and the file closes
Which hypotheses are worth testing first
Each estimate is scored on three separate things: how much the answer moves if it is wrong, how wide the honest band around it already is, and whether an experiment would narrow that band and at what cost. The tests worth running first are the ones high on all three, which is not the same as the ones everything else depends on.
| Hypothesis | Sensitivity | Uncertainty | Testability | Cost |
|---|---|---|---|---|
| Module content per unit ($6.00 injector, $2.20 cartridge) | High. It is the gate threshold and it sets Line B | High. Benchmarked to a range, not quoted | High. A customer quotation settles it outright | Under $1M |
| Recertification duration (12 months) | High. Every month removes a month of revenue at the far end | Medium. MERIDIAN has held the certificate before | High. A certification body quotes a timeline against the existing quality system | Under $0.1M |
| Merchant share of ECG electrode work (20%) | High. It carries 22% of the reachable pool | High. Unsourced, and the largest untested figure in the sizing | Medium. Two or three private-label buyers would indicate it | Under $0.5M |
| Line A price points ($0.60, $0.35, $1.20) | High. Line A is two thirds of 2032 revenue | Medium. Set below observed selling prices | High. A quotation settles them | Under $0.5M |
| Contestable share of module work (25% of merchant) | High. It is the whole basis for the module line | High. A judgment about incumbent turnover | Low in advance. Only the flow of development invitations reveals it | Not settled by the test program |
| Shared-line utilization loss | Medium. It sits inside the 9% plant charge | Medium | High. Measurable on a real line | Under $0.5M |
| CGM merchant share (5%) | Low as modeled, high as an option | High | Medium. Approaching challengers indicates it | Under $0.5M |
| Interventional share (32% of pool) | Medium on revenue, high on margin | High. Rated a long shot | Low. Only a won program settles it | Runs to 2030 |
Sequenced by how much each one tells you rather than by what depends on what:
| Order | Test | Cost | Evidence strength |
|---|---|---|---|
| 1 | Certification body gap assessment and quoted timeline | Under $0.1M | Strong. A quoted timeline is a commitment, not an opinion |
| 2 | Quotation against two module specifications | Under $1M | Strong. A price against a real specification |
| 3 | Private-label quotation against two buyer specifications | Under $0.5M | Strong |
| 4 | Approach three private-label buyers on merchant availability | Under $0.5M | Moderate. Indicates the 20% but does not fix it |
| 5 | Shared-line utilization measurement | Under $0.5M | Strong. A measurement on a real line |
| 6 | Approach three CGM challengers on second-source supply | Under $0.5M | Moderate |
| 7 | Acquisition target screen against the four partner criteria | About $1M | Moderate. A screen is not a diligence |
| 8 | Private-label electrode qualification with one customer | About $8M | Strong, and it is the one that produces revenue |
The first two cost under $1.1M together and settle the two assumptions the whole case turns on. They should run before the November 2026 decision if at all possible, because they could change the decision itself. The expensive item, qualification with a real customer, sits last deliberately: it is well understood and it should wait behind the cheap experiments that illuminate shakier estimates.
A note on evidence quality, because it decides the gate. A non-binding letter of intent is weak evidence and must not be allowed to pass for a development agreement. What counts is a document under which the customer spends its own money or commits its own engineers.
The risks no experiment can retire
These are separated from the hypotheses above because they need treatment rather than testing, and confusing the two produces mitigations that neither test nor protect.
- Treatment chosen
- Robust strategy
- Why that one
- Line A does not depend on Line B and carries the fixed organization while Line B is proven. The plan survives Line B failing; it just becomes a worse plan
- Treatment chosen
- Take the bet with eyes open
- Why that one
- This is commodity work entered knowingly. The reason to hold it is roll-to-roll utilization, not margin, and pretending otherwise would misprice the whole case
- Treatment chosen
- Robust strategy
- Why that one
- Base film runs 60% to 70% of a flexible circuit plant's operating cost from a handful of suppliers. Contract multi-year for the private-label line before committing capacity, and qualify a second source
- Treatment chosen
- Escape hatch
- Why that one
- Exposure is contained because module capacity is shared with other work rather than dedicated, so a lost program releases capacity rather than stranding it
- Treatment chosen
- Contain exposure
- Why that one
- Line C is entered last and smallest. Cap liability contractually at contract value where the customer holds the clearance. This is why the sequencing puts the highest-margin line last rather than first
- Treatment chosen
- Escape hatch
- Why that one
- The base case assumes no acquisition, so losing one costs the high scenario rather than the plan
- Treatment chosen
- Take the bet
- Why that one
- Both trends currently increase the requirement on component suppliers to be certified, which favors a certified entrant over an uncertified one. The risk is a change in direction, and nothing can be done about it in advance
- Treatment chosen
- Take the bet
- Why that one
- Ordinary manufacturing exposure, managed as the existing business manages it
Section added · Declared now because the derisking pass will write it; an addition made silently later is how an outline drifts.