Most hardware teams don't struggle gradually. They hit a cliff.
The prototype works. The pilot customers are happy. The pitch deck says TRL 7. Then the second production batch comes back with a different capacitor than the first, nobody can say which firmware build is running on the units in the field, and the one engineer who understands the provisioning script is on vacation.
The technology was ready. The company wasn't.
That is the gap the Technology Readiness Level scale doesn't measure. TRL grades the product. It says nothing about whether your organization can build it twice the same way, change it without breaking it, or support it once it's in a customer's hands. For an IoT startup, that second question is the one that kills you.
TRL Measures the Product. Nobody Is Measuring the Company.
NASA developed TRL to grade technology maturity for space programs. It has since spread to the Department of Defense, EU research funding, and plenty of investor diligence checklists. It asks one question: does this technology work, and in what environment has that been shown?
It's a good question. It's just not the only one. TRL doesn't ask who builds the product, from which BOM revision, or what happens when you change a part.
The Department of Defense ran into this directly. A working technology turned out to be a poor predictor of a smooth production ramp, so DoD added a second scale: Manufacturing Readiness Levels, MRL 1 through 10. MRLs grade what TRL ignores: design stability, materials and supply chain, process control, quality management, workforce, and manufacturing management. A defense program isn't considered ready for low-rate initial production until MRL 8, meaning a pilot line has been demonstrated, not just a working prototype.
You don't need the MRL deskbook. It's written for defense acquisition programs, and it reads like one. The principle transfers anyway: readiness runs on two tracks, and the slower track sets your real level.
Two Tracks, One Ladder
Below is the TRL ladder adapted for IoT, with a second track beside it: what your organization has to be able to control at that level. I've also mapped each level to the build stages most hardware startups actually run, EVT, DVT, and PVT.
The mapping is approximate. TRL measures evidence; build stages are schedule milestones. They don't line up perfectly, but they line up well enough to plan with.

TRL 1–3: Concept and Feasibility (Before EVT)
The product: The core idea is validated on paper and on the bench. Key components are identified: MCU, radio, sensors, power. You have early firmware experiments and bench-level connectivity tests, but nothing is integrated yet.
The organization: Almost nothing, and that's correct. Process at this stage is overhead on a design that's supposed to change daily.
The one habit worth starting now: write down why you picked the parts you picked. A notes doc is enough. Six months from now, someone will ask why the board uses that radio module, and "it was on the dev kit" is not a reason you want to rediscover.
TRL 4: Integrated Prototype (Before or at the Start of EVT)
The product: Your first integrated prototype. A custom PCB or a dev-board stack, with firmware, radio, and sensors working together. Cloud ingestion works under controlled conditions. Flashing and configuration are manual.
The organization: You need to be able to answer "what's on this board?" for every unit on your bench. That means part numbers and one BOM per build. This isn't a PLM rollout; it's a part numbering scheme and a single place the BOM lives.
This is where many startups raise their first serious round, and the demo makes the product look closer to done than it is. Investors see TRL 4. Your organization is at zero.
TRL 5: Relevant Environment (EVT)
The product: Devices leave the lab for early field trials. Connectivity varies, power consumption problems surface, and edge cases start dominating behavior. Failures become intermittent and harder to reproduce.
The organization: This is where the design becomes worth preserving. Once a working build is out in the field, every change needs to be tracked against a known baseline, or the next intermittent failure becomes untraceable. EVT is when to turn on change orders: lightweight, but real.
You also need to know which unit is which. Every serial number should map to a hardware revision and a firmware build. Observability matters for the company as much as for the device.
TRL 6: System Demonstration (DVT)
The product: Pilot deployments with real users. Your contract manufacturer is building test units. Multiple devices are in the field, you've attempted your first OTA updates, and the support burden has started.
The organization: The system works here, but usually because the team is compensating by hand. That is an organizational gap, even if it shows up as a technical symptom.
Your CM is now building from your data, so the BOM they receive has to be the BOM you mean. A shared spreadsheet gets through this stage on luck, and this is usually where teams discover the spreadsheet has stopped being enough. Firmware also has to become a versioned item tied to a hardware revision, not a hex file in someone's Downloads folder. Firmware belongs in the BOM for exactly this reason.
TRL 7: Operational Prototype (PVT)
The product: Dozens to hundreds of units in operation, depending on your use case. Manufacturing variability shows up. Firmware versioning becomes critical. Support and ops load climbs sharply, and compliance questions start blocking sales.
The organization: This is the cliff. Every change now touches your CM, your inventory, and units already in the field. Your organization has to be able to do three things it could skip before: hand your CM a complete package, not just a BOM; check the where-used impact of a change before making it; and say exactly what's deployed, by revision.
PVT is the latest point to learn change control. Learning it here means learning it while it's load-bearing.
TRL 8: Qualified System (Production Ramp)
The product: Production-intent hardware with a defined manufacturing process and repeatable provisioning. Certification is complete or underway.
The organization: Speed now depends on coordination, not engineering talent. A certified design is a frozen design; a change to the radio, the power stage, or the enclosure can reopen certification. Your organization needs to know which pending changes do that before they're approved, not after the test lab tells you.
This is also the level where knowledge that lives in one engineer's head becomes a production risk. If only one person can release a build, your production line has a single point of failure, and it isn't a part.
TRL 8 is the rough IoT equivalent of the MRL 8 bar: you have demonstrated you can build it repeatably, not just that it works.
TRL 9: Proven at Scale
The product: Hundreds to thousands of devices deployed. Stable manufacturing and supply chain, predictable updates, measurable reliability.
The organization: At this level the two tracks merge, because reliability at scale is a property of your process as much as your design. Failures are costly, visible, and reputational. When one happens, you need to trace it back to the exact build, BOM revision, firmware version, and the change that introduced it.
The Ladder on One Page
| TRL | Build stage | Product evidence | Organization must control |
|---|---|---|---|
| 1–3 | Before EVT | Feasibility shown on the bench | Rationale for key part choices |
| 4 | Before or early EVT | Integrated prototype works in the lab | Part numbers; one BOM per build |
| 5 | EVT | Works outside the lab in field trials | Change orders against a baseline; serial numbers mapped to hardware and firmware revisions |
| 6 | DVT | CM-built pilot units with real users | A BOM your CM can build from without you; firmware versioned per hardware revision |
| 7 | PVT | Dozens to hundreds of units in operation | Complete CM handoff package; where-used impact of every change; deployed units by revision |
| 8 | Production ramp | Qualified, certified, repeatable build | Change control that flags certification impact; no single-person knowledge |
| 9 | Scale | Proven reliability in operation | Any field failure traceable to build, BOM revision, firmware, and change |
Your Real TRL Is the Lower of the Two
When an investor or an enterprise customer asks what TRL you're at, they aren't really asking whether unit 5 works. They're asking whether unit 500 will match it.
A product at TRL 7 run by an organization at TRL 4 is a TRL 4 company with a good demo.
If you can't answer these questions, your organization is behind your product:
- Which hardware revision and firmware build is on each unit in the field?
- If your CM substituted a part in the last batch, would you know?
- What changed between your last two builds, and who approved it?
- If your lead hardware engineer left tomorrow, could someone else release the next build?
- Which pending changes would reopen certification?
None of this requires an enterprise PLM rollout. It requires the organizational track to climb at the same pace as the product, which for an IoT startup means one place for mechanical, electrical, and firmware data. That is the gap oroForge is built for: part numbers, BOMs, and change orders sized for a team climbing from EVT to production, with a free plan and a flat-rate Team plan to start on.

