Do not sign off a custom ebike preproduction sample on a photo, a battery certificate, or a supplier's verbal assurance. I approve only when three things point at the same bike: the frozen build, the accredited laboratory's certificate scope, and a matched battery-charger-system record. One condition flips that judgment. If your purchase contract sets a different evidence standard, the contract governs, and if the sample in your hands is not the configuration you will actually sell, the file is answering the wrong question.
Freeze the Build Before Signoff
A signoff without a build list approves nothing in particular. Custom work is exactly where certificate coverage gets thin, because frame size, motor, pack, charger, controller, display firmware, and assembly state each move the product away from whatever was tested. Compared with a stock model, a custom build drags more variables into the sample, which is why the configuration sheet carries the weight here. Write those items into one sheet with a revision number and a date, and get the supplier to sign that revision.
The condition you approve has to be the condition you sell. Under 16 CFR 1512, a bicycle subject to the regulation must meet the requirements in the condition offered for sale, and a bike offered in partially assembled condition must meet those requirements after assembly according to the manufacturer's instructions. Read that against a normal container: the sample arrives fully built, the carton arrives with the front wheel off and the bars turned. If assembly state never enters the configuration sheet, nobody has approved the thing that reaches the customer.
So I put an assembly-state line and an instruction-sheet version on the sheet. Anything absent from the sheet stays unapproved, and a mismatch caught at the desk costs less than one caught in a warehouse.
Check Certificate Scope Against the Build
A certificate describes a product identified on the paper; it is not a statement about a factory. The CPSC letter to micromobility stakeholders is direct about the form the demonstration should take: compliance with the standards should be demonstrated by certification from an accredited testing laboratory. What that sentence does not do is stretch a certificate onto your bike.
Read the paper the way a customs broker reads an invoice: holder name, the lab's accreditation, product and model identifiers, the component list, the standard and its edition, and whether the pack and charger appear as tested items. Compared with a packet that lists the charger as a covered component, a battery-only certificate leaves the pairing outside the paper. Unlike a generic battery certificate, a build-specific trace names the frame, pack, and charger you are actually buying.
The letter is guidance that carries a 2022 compliance call, not a ruling that every standard it mentions is nationwide mandatory law, and applicability differs across the standards it discusses. Confirm which standard covers your configuration with your own compliance reviewer. The scope check here asks one narrow question: does the document talk about this build?
Audit Battery and Charger as One System
A battery certificate on its own is weaker evidence than it looks, because the electrical parts succeed or fail as a set. CPSC describes the practice as a system approach ensuring that the device, battery, cells, and charger are compatible and function safely together. One record, four part identities, one condition of approval.
Suppose I run that audit at a desk with no bench and no test rig. I lay the configuration sheet beside the certificate annex and a photo of the nameplate, then build a trace listing the bike model and serial range, the pack part number, the charger part number, and the annex reference for each. Every line gets marked matched, missing, or outside scope. What comes out is a document map, not an electrical result, and I would not describe a matched trace as passed.
The mismatches are the payoff. A charger that never appears in the annex. A pack revision that postdates the certificate. A pack from one supplier married to a charger from another with no joint statement covering the pair. Each one is cheaper to catch before the deposit than after the container lands.
Red-Team Certificate Version and Model Limits
Assume the supplier sends a clean-looking certificate. I try to break it before I trust it. Three moves do most of the work: compare the certificate date and standard edition against the documents already on file, compare the model, pack, charger, and assembly state against the configuration sheet, and flag every point where the paper predates a change.
Any of these flips a signoff to hold: an undated certificate, a component that appears in the build but not in the annex, a certificate issued before an engineering change recorded on the sheet, or a model suffix nobody explained. Hold means the sample sits until a document covers the current build, and a deadline does not convert missing paper into coverage.
Version discipline cuts both ways. 16 CFR 1512 as printed in the 2025 annual CFR volume is a snapshot of that edition, and an annual compilation does not promise that later amendments are folded in. Certificates age the same way. A standard displayed on a webpage is not proof of what your bike was tested to, and a certificate downloaded this year says little about a build changed last month unless the identifiers still line up.
Run a Hypothetical Signoff Evidence Matrix
Imagine a one-page matrix filled in before anyone signs. Rows carry the configuration sheet, the certificate-to-build match, the battery-charger-system trace, a nameplate and photo cross-check, the assembly-state statement, and the open RFQ questions. Columns record whether the document is present, whether it matches the build, whether a scope gap remains, and the decision sitting at the end of each row: hold or approve.
The only pass rule I propose is documentary completeness for the exact build: no performance threshold, no test result, no claim that a matched trace means the bike is good. Both the sample and the production build must point to the same sheet revision for that rule to mean anything. Every cell stays pending until supplier documents land.
That is also where most packets fall short: until the documents arrive, half those cells are not on the table yet. A blank scope column is not a maybe. If it is blank, hold the pen and do not sign, because the matrix only works when empty means empty.
Ask These RFQ Questions Before Sampling
Put the questions in writing and ask for documents rather than assurances. Which exact configuration and assembly state ships, and which configuration sheet revision covers it? Who holds the certificate, which accredited lab issued it, which models and components does it name, and what standard edition and date does it carry? Which pack part number and which charger part number were evaluated together, and where is that pairing recorded? What changed since the certificate was issued, and is there a change log?
A supplier who cannot name the accrediting lab, the pack part number, or the sheet revision is telling you where the file stands. That answer is useful even when it is unwelcome, because it arrives while you can still renegotiate the sample instead of after the tooling is cut.
Keep the questions inside the evidence. MOQ, price, lead time, and tooling commitments are separate conversations, and asking a supplier to fill those blanks inside a signoff document invites numbers nobody can support later. Nothing in this file replaces lot inspection, your own compliance review, or a lawyer's read of the contract.
This is not for buyers who need a certificate to double as a build review. It is for teams that will hold the sample until the configuration sheet, certificate scope, and matched battery-charger trace point to the same revision; the next step is to request that document set and log every gap as a hold.
