Legal steps to protect intellectual property for startups: the order that actually saves you
A founder once asked me to review her cap table two weeks before a seed round. Great numbers, clean deck, three committed investors. Then the lead investor's lawyer asked a single question: "Who owns the code your CTO wrote before incorporation?" Nobody knew. The round closed four months late, at a lower valuation, after a retroactive assignment agreement and one very uncomfortable conversation with a former contractor. That question cost them roughly €180,000 in dilution and a quarter of momentum.
Intellectual property for startups isn't a legal formality you handle once. It's a sequence, and getting the order wrong is what hurts. Most founders think IP means "file a patent." It doesn't. The legal steps to protect intellectual property for startups start way before any filing office is involved — and the first step costs nothing but a signature.
Key Takeaways
- IP ownership comes first. If your company doesn't own the work, no filing will save you.
- Every founder, employee, and freelancer needs a written assignment clause. No exceptions.
- Trademarks are the cheapest protection with the fastest return. File early, in your core class.
- Patents are slow and expensive — decide based on whether your moat is technical or executional.
- Trade secrets require operational discipline, not legal paperwork: NDAs, access control, documentation.
- Investors run IP due diligence on ownership chain, not on glamour filings.
Why most startups mishandle IP (and it's rarely laziness)
The problem isn't that founders don't care. It's that IP work has no deadline. Nobody churns because you haven't assigned your trademark. Your product still ships, users still sign up, revenue still grows. So it slides.
Then a trigger arrives — a fundraise, an acquisition offer, a cease-and-desist letter, a co-founder leaving — and suddenly the missing pieces become expensive. I've watched this play out maybe a dozen times. The pattern is identical every time.
The three real failure points
Here's what actually goes wrong, in order of frequency:
- Unsigned assignment clauses. A contractor built your MVP. The contract said "deliverables" but never transferred copyright. Under most legal systems, the author keeps the rights unless explicitly assigned.
- Premature disclosure. A pitch at a demo day, a detailed blog post, a conference talk. In many jurisdictions, novelty is destroyed by public disclosure, and your patent window closes permanently.
- No internal record. You can't prove a trade secret existed if there's no timestamped documentation of what was secret and who had access.
Notice that none of these are about filing offices. All three are about paperwork you could have signed on day one.
The sequence: what to do, in what order
Legal protection isn't a list of options. It's a timeline. Here's the order I'd defend in any jurisdiction.
Phase 1 — before you launch (weeks 1 to 4)
This phase is entirely about ownership. Nothing else matters yet.
- Founder IP assignment agreement. Every founder assigns all relevant past, present, and future work to the company. This covers code written on a laptop before incorporation — a detail people forget constantly.
- Employment and contractor clauses. Include a written assignment of all work product, a confidentiality obligation, and a non-solicit where enforceable. For freelancers, add an explicit waiver of moral rights if your jurisdiction recognizes them.
- A simple invention log. A dated document where technical decisions and novel methods get recorded. It's crude, but it establishes a timeline.
Total cost at this stage: legal template review, typically a few hundred euros or dollars. Total value: it's the difference between a fundable company and a liability.
Phase 2 — from launch to first meaningful revenue (months 1 to 12)
Now you file. Selectively.
Trademarks come first, almost always. Your name, your logo, and — if you're selling a product — the product name. A trademark filing in a single class in the EU runs roughly €850 for a standard application covering one class, with additional classes costing a couple hundred more each. In the US, expect filing fees in the low hundreds of dollars per class, plus attorney time if you use one. Registration typically takes several months, but your protection generally dates back to the filing.
The class question trips people up. You don't need all 45. You need the classes where you actually sell today, plus one or two where you plausibly will within a few years. I've seen startups burn thousands on defensive classes they never used.
Copyright is automatic in most jurisdictions, but registration still has value in some countries — notably the US, where timely registration enables statutory damages in infringement suits. That's a real lever, not a formality.
Phase 3 — patents, trade secrets, and the fundraise
Patents are where founders waste the most money, because they file on reflex. Ask one question instead: can a competitor copy my core advantage by reading my product?
If yes, and the advantage is a technical mechanism, a patent (or a provisional filing to buy time) may be worth it. If your advantage is speed, brand, network effects, or operational know-how, a patent is an expensive distraction. Full utility patents routinely cost five figures once prosecution and attorney time are included, and examination often stretches well beyond two years.
For everything that isn't a patent candidate, you're relying on trade secrets. That's a legal category, but its strength is operational. Which brings up the practical side.
Trade secret discipline: the unglamorous part
Four things make a trade secret defensible:
- NDAs with anyone who sees the internals — investors, contractors, potential partners. A mutual NDA is standard and takes minutes.
- Need-to-know access. Not everyone on the team needs the full architecture or the margins.
- Documented policies. An internal note stating what's confidential is worth more than you'd expect in a dispute.
- Exit hygiene. When someone leaves, revoke access the same day. I've seen a departing engineer walk out with a full repository on a personal drive because nobody thought about it until Monday.
None of this is legal work in the traditional sense. It's management. And yet it's what most trade secret disputes turn on.
Comparing your protection options
| Protection type | What it covers | Typical cost | Time to effective protection | Best fit |
|---|---|---|---|---|
| Trademark | Brand names, logos, product names | Low to moderate per class | Months (rights from filing) | Any company with a public-facing name |
| Copyright | Code, content, design files | Free, registration optional | Immediate on creation | Software, media, written content |
| Patent | Novel technical mechanisms | High, ongoing | Years | Deep tech, hardware, biotech |
| Trade secret | Know-how, processes, data, algorithms | Operational, not legal | Immediate, if disciplined | Most SaaS and service businesses |
| Assignment agreements | Ownership chain itself | Very low | Immediate on signature | Every startup, no exception |
That last row is the one founders skip. It's also the only one that, if missing, can invalidate your entire position.
Jurisdiction changes the steps more than most guides admit
Nearly every IP checklist you'll find online is written from a US perspective. USPTO, statutory damages, provisional applications. Useful, but incomplete if you're operating elsewhere.
In the EU, you have the European Union Intellectual Property Office for a single trademark covering all member states, plus national offices like the INPI in France or the DPMA in Germany for narrower, cheaper filings. Copyright arises automatically across the bloc. Software patents are harder to obtain than in the US, which pushes more European startups toward trade secret strategy by default — sometimes without realizing it.
If you plan to sell internationally, the Madrid System lets you extend a trademark across many countries from a single application. For patents, the PCT route buys you time to decide which national offices to enter. Both are mechanisms for deferring cost, not avoiding it.
My advice: file first where you have customers and where your name is most at risk of being squatted. Then extend as revenue justifies it.
What investors actually check
When a fund's legal counsel opens your data room, they aren't counting your patents. They're tracing ownership.
The standard request list looks roughly like this: all founder and employee assignment agreements, all contractor agreements with IP clauses, a schedule of registered IP with filing numbers and status, any licensing agreements, and a description of any known disputes or third-party claims. They also ask whether any IP was developed using university resources, previous employer equipment, or government funding — three classic sources of hidden claims.
I once sat in a diligence call where the acquirer's lawyer spent forty minutes on assignment paperwork and about four on the patent portfolio. That ratio tells you everything about what's actually being valued.
If you're raising in the next twelve months, do this audit yourself now. Finding a gap is cheap. Having an investor's lawyer find it is not.
Mistakes that cost the most
Ranked by damage, not by how often they're mentioned:
- Assuming the company owns founder work. Incorporation doesn't retroactively transfer anything.
- Filing a trademark after a public launch. Squatters monitor launch platforms. This is not paranoia; it's a business model for some people.
- Treating NDAs as sufficient. An NDA without restricted access to the secret is weak evidence.
- Filing patents for investor optics. Sophisticated investors see through it. You spent six figures on a shield you don't need.
- Forgetting international exposure. Your name may already be registered elsewhere by someone faster.
Number one is the one that ends deals. Everything else is expensive. That one is fatal.
If you do only one thing this week
Write down every person who has touched your code, your designs, your brand assets, or your customer data — founders, employees, freelancers, agencies, interns. Then check, one by one, whether a signed document transfers their work to the company.
If the answer is "I think so" or "it's in the contract somewhere," that's your project for the next seven days. Not the patent search. Not the trademark class analysis. The ownership chain.
Because here's the thing about IP for startups: the filings are the visible part, the part you can photograph for a board deck. The invisible part — a signature on a page, a date, a clause — is what determines whether any of it was ever yours to protect. And the strange part is that the invisible work takes a week, while the visible work takes years.
Most founders get that ratio backwards. Don't be most founders.
