Release a hardware tranche against physical evidence, not against a date. The verifiable manufacturing milestones are tool release on a first article inspection report, a capability run at production cycle time, a scrap rate that holds steady at or below the figure in the cost model, and a unit cost confirmed by a landed invoice rather than a quotation. Each of those produces a document that an independent engineer can check, which is what makes a milestone usable, and what software style milestones struggle to deliver in a hardware context.
Why do standard milestone definitions fail on hardware?
Because most published frameworks were written for products that ship as code or pass through a regulator. Beta release, user numbers, a clinical endpoint and a regulatory submission are all verifiable, and none of them tells you whether a factory can make 50,000 units at the cost in the model.
Hardware fails in a different place. A company can have a finished design, a signed distribution agreement and passed pre-compliance testing on prototypes, and still be two tooling corrections and four months away from a part that comes out of the mould within tolerance. The capital in a hardware round is overwhelmingly spent on physical readiness: tooling, qualification, first inventory. If the tranche conditions do not measure physical readiness, they are measuring something else.
There is a useful public reference point here. Manufacturing Readiness Levels, a 1 to 10 scale developed by the United States Department of Defense to describe how mature a manufacturing process is, run alongside the better known Technology Readiness Levels. They are a reasonable frame for a term sheet discussion, but they are deliberately broad. For a tranche condition you need something a toolmaker can hand over in an email, and that is the list below.
Which manufacturing milestones are objectively verifiable?
Eleven, in the order they occur. Each one produces a specific document, which is what makes it usable as a release condition rather than a conversation.
| # | Milestone | Evidence it produces | | --- | --- | --- | | 1 | Design frozen and reviewed for manufacture | A design for manufacture report with findings accepted or declined, against a model revision | | 2 | Tool design approved | Cavity count justified against the volume in the business plan, cooling and steel grade specified | | 3 | Tool ownership documented | A purchase agreement naming the company as owner with a right to move the tool | | 4 | T1 samples measured | A full dimensional report against the drawing, not a photograph of a part | | 5 | Tool released | A signed first article inspection covering every drawing dimension, on parts from each cavity | | 6 | Capability established | A run of roughly 300 consecutive parts at production cycle time, with a capability index against the critical dimensions; 1.67 is the usual acceptance level for a new process under PPAP, with 1.33 a common minimum for ongoing production | | 7 | Scrap rate stable | Scrap per cavity across three consecutive runs, at or below the cost model figure | | 8 | Certification in hand | The certificate or declaration of conformity for the target market, with test reports on parts from the production tool | | 9 | Unit cost verified | A landed invoice for a representative shipment, reconciled line by line against the cost model | | 10 | Capacity confirmed | Weekly output at current tooling, with contracted running hours, set against the sales plan | | 11 | Second source scoped | A named alternative with a costed qualification path and lead time |
Milestones 5 and 6 are the pair most often collapsed into one, and they are not the same event. A first article inspection establishes that the tool can produce a conforming part. A capability run establishes that the process does so repeatably. Companies can pass the first and fail the second by a wide margin, and the gap between them can be a tooling correction and six to ten weeks.
How should those milestones map onto the tranche schedule?
In four releases, each one unlocking the spend that the previous evidence justifies. A workable structure:
1. Tranche 1, on milestones 1 to 3. Funds tooling. The evidence is a manufacturability report, an approved tool design matched to the planned volume, and ownership on paper. Releasing tooling capital without milestone 3 is one of the most avoidable losses in hardware investing: the money is spent, the asset exists, and the company cannot move it. 2. Tranche 2, on milestones 4 and 5. Funds qualification, the first material purchase and packaging. Evidence: a dimensional report and a signed first article inspection. 3. Tranche 3, on milestones 6 to 8. Funds the first commercial inventory. Evidence: a capability run, a stable scrap rate and the certificate on production parts. 4. Tranche 4, on milestones 9 to 11. Funds scale up. This is the tranche where the cost model stops being a model, because milestone 9 replaces the quoted piece price with a landed cost from a representative commercial shipment. Further evidence: documented capacity against the sales plan, and a second source scoped so that one factory, one tool and one shipping lane are not all single points of failure.
Build slippage into the schedule rather than pretending it will not happen. Tool build alone runs six to twelve weeks, one correction round after T1 is normal and two is unremarkable, and each round adds two to four weeks. A tranche schedule that assumes zero correction rounds will be renegotiated, and a renegotiation under time pressure is expensive for both sides.
Which milestones are most often left out?
Four, and all four show up as unbudgeted cash later.
- Tool ownership. Paying for a tool does not automatically make you the owner of it. Ownership follows the contract wording. Where it is unclear, a factory change can arrive with a transfer charge that was never quoted, and in some arrangements the tool simply does not move.
- Verified landed cost. A piece price quotation routinely excludes packaging, inbound freight, duty, scrap allowance, outgoing inspection and any secondary operation. Those lines are frequently a double digit percentage of unit cost. Conditioning a tranche on a reconciled landed invoice removes the single largest source of error in a hardware cost model.
- Certification evidence in hand. Not "testing scheduled" but the certificate or declaration of conformity for the market being sold into, backed by test reports, on the exact product configuration coming out of the production tool. Design changes made during tooling correction can invalidate testing done on prototypes. [Higo Sense](/projects/higo-sense) had a one year certification deadline, on a device with more than sixty parts from thirteen moulds.
- Working capital for the gap between paying and selling. Material and tooling are paid months before the first invoice is collected. The first production run is typically the largest single cash outflow in the entire plan, and it lands at the point where the schedule is most likely to have slipped.
One trade off is worth naming, because it is real and it cuts both ways. Tight milestone conditions reduce the investor's exposure and shift timing risk onto a company whose schedule depends on a toolmaker it does not control. Set the bar so high that a normal correction round triggers a default, and you will spend the next quarter renegotiating instead of building. The useful conditions are the ones that are objective and achievable: a signed inspection report, a measured scrap rate, a landed invoice, a documented owner.
[Production Heroes](/) takes products from design to series production, from feasibility and cost price to sourcing and supply chain, and supplies moulds and tooling to manufacturers across Europe, from the Netherlands and its own presence in China.
Sources: Manufacturing Readiness Levels as published by the United States Department of Defense and summarised in public engineering references; dimensional results per cavity, capability run size and capability index thresholds from PPAP (AIAG), the widely used industry standard for production part approval. Durations are practical ranges, not commitments.
Frequently asked questions
- What proves a mould is production ready?
- A first article inspection measuring every drawing dimension on parts from the production tool, followed by a capability run at production cycle time. The inspection shows the tool can make a conforming part; the capability run shows the process repeats it. Release the toolmaker's final payment on the first.
- What scrap rate is acceptable in a first production run?
- Expect elevated scrap while the process window is found, and judge the trend rather than the first number. For a tranche, what counts is whether the rate holds steady at or below the figure in the cost model across three consecutive runs.
- How long does first article inspection take?
- One to three weeks from sampling to a signed report, depending on how many dimensions the drawing carries and whether measurement is done in house or by a laboratory. The delay is rarely the measuring; it is that nobody specified which report and who signs.
- What does a capacity check look like?
- Cavity count times 3,600, divided by cycle time in seconds, times available running hours. That is the theoretical maximum: correct it for scrap and downtime before comparing with the sales plan. Ask how many hours are contractually yours, because a press shared with higher margin work is not reliable capacity.
- Should tooling ownership be a condition of a tranche?
- Yes. The tranche funding tooling is buying a physical asset, and if the purchase agreement does not name the company as owner with a documented right to move the tool, the investment has bought a production dependency instead. Verify the contract wording, not the invoice.
- How reliable is a COGS estimate before the first production run?
- Treat anything before a landed invoice as a model. Piece price quotations routinely exclude packaging, inbound freight, duty, scrap allowance, outgoing quality control and secondary operations. Those lines frequently add a double digit percentage, so verify unit cost against a real landed shipment before the scale up tranche.