Key takeaways
- A portfolio page is marketing copy with pictures attached. Nothing stops a firm displaying work it only partly did.
- Renders, prototypes and shipped products look similar in a photo and mean very different things about capability.
- A firm’s most impressive project is often the one it cannot show you, because confidentiality restricts disclosure.
- A reference the firm supplies is a floor, not a verdict. It chose that person because the conversation would go well.
Checking a development firm’s portfolio and references means separating what the firm chose to show you from what you can verify on your own. A portfolio page and a glowing testimonial cost a firm nothing to produce. A reference who answers specific questions honestly, and a product you can independently confirm actually shipped, cost something: time, and the risk that the answer is not flattering. That distinction, marketed evidence against verified evidence, is what this guide is built around.
This is one piece of a larger vetting process. For the full six-stage sequence from shortlist to signature, see our guide to how to vet a hardware development company before you sign.
Why portfolios are the easiest thing to fake
A portfolio page is marketing copy with pictures attached, and nothing stops a firm from displaying work it did not do, work it did only a small part of, or renders that were never built. The Federal Trade Commission’s Endorsement Guides require that any endorsement reflect the honest experience of the endorser, with adequate substantiation behind the claims it makes. That rule exists because unverified praise is cheap to produce and easy to misuse.
- A CAD render can be mistaken for a finished product by anyone who does not build hardware for a living.
- A “case study” can describe a project the firm advised on briefly, not one it carried from concept to production.
- A client logo on a website does not confirm the scope of what that client actually paid for.
Renders, prototypes, and products that actually shipped
How to tell which one you are looking at
These three categories look similar in a photo and mean very different things about a firm’s capability. A render proves the firm can operate design software. A prototype proves a design existed physically at some point. Only a shipped product proves the design survived tooling, certification, and a production line.
| What you see | What it actually proves | How to tell |
|---|---|---|
| CAD render | The firm can model a concept | Perfectly clean surfaces, no visible seams, parting lines, or screw bosses |
| 3D-printed or machined prototype | A design existed physically at some point | Visible print layers or machining marks, often a single unit, no packaging |
| Shipped product | The design survived tooling, certification, and volume production | Retail packaging, a listing on a retailer’s site, a visible regulatory mark, or a serial number you can trace |
Ask directly: is this a render, a prototype, or something that shipped? A firm with a real track record answers that question immediately, for every item in its portfolio, without needing to check with anyone first.
Why the best work is often under NDA
A firm’s most technically impressive project is often the one it cannot show you. Confidentiality agreements routinely restrict disclosure of the client’s identity, the product’s specifications, and sometimes the existence of the engagement at all. A firm that never mentions confidentiality is either working only on unremarkable projects or is not being straight about what it can share.
What a firm can show you without breaking confidence
- Generic process work: a DFM review, a test fixture, a tooling sample, with client-identifying details removed.
- Category-level descriptions: “a connected home device,” “a wearable medical accessory,” without the client’s name or exact specifications.
- References who can speak to how the firm worked, even if the reference cannot describe what was built.
A firm that shares little but still offers a call with a vouching client is behaving consistently with a real confidentiality obligation. A firm with nothing to show and no one to vouch for it either is a different situation.
Reading a case study critically
A case study is written by the firm about itself, so treat it as a claim to verify, not a fact to accept.
- Does it name the specific problem the client had, or only describe a generic process?
- Does it name what went wrong along the way? A case study with no setback, on a first-of-a-kind hardware product, is unusual enough to be worth asking about directly.
- Does it specify which disciplines the firm handled itself (mechanical, electrical, firmware, industrial design) and which were subcontracted?
- Does it name a result you could independently check, such as a retail listing, a certification grant, or a production volume?
The same discipline applies to how a firm behaves once you start asking these questions. A firm that deflects rather than answers specifically is showing a pattern covered in more depth in our guide to red flags in a product development partner.
Getting real reference calls
A reference the firm supplies is a floor, not a verdict; the firm chose that person because the conversation would go well. Getting a useful reference call is less about who you ask and more about what you ask once you are on the call. Our companion list of questions to ask before hiring a hardware development company covers the broader interview; the questions below are specific to the reference call itself.
Asking for a reference on a project that went badly
Ask the firm directly for a reference on an engagement that did not go smoothly, not only its best one. How a firm answers that request is itself informative: a firm that names one and explains what happened is behaving differently from a firm that insists every project has gone perfectly.
Questions that get honest answers
- “What happened when the schedule slipped, and who told you first?” A specific answer describes a real event. A vague one does not.
- “What did the first prototype fail on, and what changed after that?” Common products with established functions and few unique features can pass the first build without changes.
- “If you had to do this again, would you use the same firm?” Ask this last, after the specific questions, so the answer is grounded rather than reflexive.
- “Was there a cost or schedule surprise, and how was it communicated?” This tests how the firm handles bad news, not just good news.
Back-channel checks you can do yourself
Some verification does not require the firm’s cooperation at all.
- Search the firm’s name with “complaint” or “scam” added to the query. The Federal Trade Commission’s own guidance for small businesses recommends this as a basic first step before doing business with an unfamiliar company.
- Check the FCC’s public equipment authorization database if the firm claims to have taken a wireless product through certification. FCC grants are tied to a named grantee, so a claimed certification project involving radio hardware should leave a public trace you can search yourself.
- Check the state where the firm is registered to confirm the entity exists and is in good standing, not suspended or dissolved, before you rely on any claim it makes about itself.
- Ask people you trust for a recommendation rather than relying only on inbound sales contact. This is the same basic guidance the FTC gives consumers evaluating any unfamiliar vendor.
When thin evidence is still acceptable
Every legitimate firm was new once, and a thin portfolio is not automatically disqualifying. What matters is whether the firm is honest about the gap. A newer firm with strong named engineers, one real reference willing to talk in detail, and a clear account of what it has and has not done is more defensible than an established firm that will not let you speak to a client at all. The problem is never a short track record alone. It is a firm that behaves as though the gap does not exist.
Frequently asked questions
How many references should I actually call?
Two or three, if you can get past the first name the firm offers. The first tells you the firm has one happy client. The second and third tell you whether that is typical.
Is it reasonable to ask for a reference from a project that failed?
Yes. Every firm with any real history has had a difficult engagement. Asking for that reference, and watching how the firm responds, tells you more than another success story would.
What if a firm’s best work is genuinely all under NDA?
That is common and not itself a red flag. The distinction is whether the firm can still offer generic process examples and one live reference, or offers neither and asks you to trust its claims.
Should I be suspicious of a firm with only renders in its portfolio?
Ask directly whether any shown work reached production. A newer firm may have more prototypes than shipped products. What matters is whether the firm is clear about which is which when asked.
Can I verify a claimed certification project myself?
For wireless products, yes, through the FCC’s public equipment authorization database, since certifications are tied to a named grantee. For other certification types, ask the firm which lab or certification body was used and confirm that body’s accreditation independently rather than taking the firm’s word for the result.
What is the single best question to ask a reference?
“What happened when something went wrong, and how did the firm handle it?” A specific, concrete answer is worth more than any number of positive but vague responses.
Where Inventornest fits
Inventornest expects this kind of scrutiny and treats it as a normal part of a founder’s diligence rather than an inconvenience. We can describe what is and is not under NDA on past work. If you want to run this exact process against a firm you are evaluating, or against us, you can book a consultation. Our OEM development and manufacturing services page has more on how we scope and document engagements.
