I had a founder on a call last Tuesday — [redacted seed-stage AI infra founder] — and the first thing he said was: "We filed a provisional six months ago. Are we covered?" I asked him what he had inside it. He paused. Then he said, "I think it was a slide deck."
That conversation is the reason I am writing this. Because the provisional patent is, in my experience, the most misunderstood instrument in startup IP. Founders treat it like an insurance policy when it is closer to a placeholder. They file it to feel protected, then carry on building as if the work is done. It is not done. The provisional is a starting line, and most founders never run the race that follows.
I have personally reviewed somewhere north of 200 startup patent filings across pre-seed through Series B companies. The pattern is brutally consistent. The founders who treat the provisional as a strategic option turn it into real defensibility. The ones who treat it as a finish line burn between £1,500 and £3,000 and end up with nothing — or worse, with a public disclosure that destroys their patentability everywhere else.
So here is the honest answer. Yes, a provisional patent can be worth it. But the conditions matter more than the filing itself.
What the provisional actually gives you
The provisional buys three things, and only three things. A priority date. The right to say "patent pending." And a twelve-month window to decide what to do next.
The priority date is the only one of those three with real legal weight. It anchors your invention to a specific moment in time. If a competitor files something similar after you, your earlier date is what determines who has the first claim. In sectors moving as fast as AI, that head start can decide whether you own a position or read about it on someone else's patent grant.
The "patent pending" label has commercial weight, not legal weight. It signals to investors, customers, and acquirers that you take IP seriously. That signal matters. Companies with patents are materially more likely to close their early rounds. But the signal only holds if you actually file the utility patent within twelve months. If you do not convert, "patent pending" disappears, and you are back to where you started — except you have already disclosed your invention.
The twelve-month window is the part most founders waste. It is a research and validation runway. You can test market fit, refine your claims with real customer feedback, narrow what you actually want to protect. The founders who use this window well file utility patents that are sharper, narrower, and far more defensible. The ones who treat it as nap time file the same vague description twelve months later and wonder why the examiner shreds it.
When the provisional is genuinely worth it
I tell founders the provisional is worth filing in three situations, and I am sceptical of it in most others.
The first is when you are about to disclose. If you are pitching investors who will not sign an NDA, demoing at a conference, publishing a research paper, or open-sourcing part of your stack — anything that creates a public disclosure — the provisional protects your priority date before the world sees the invention. Without it, that disclosure can permanently bar you from filing in most jurisdictions outside the US.
The second is when you are racing a credible competitor. If there is a known player in your space who could file something overlapping, the provisional locks in your date for the cost of a decent laptop. That is cheap insurance against being second to file on something you invented first.
The third is when the invention is genuinely unfinished. If you have the core idea but you are still iterating on the implementation, a well-drafted provisional preserves the priority date while you finish the work. The keyword there is "well-drafted" — I will come back to that.
What the provisional is not worth doing is filing one because a friend told you to, because it sounds professional, or because you want a line on your pitch deck. None of those justify the spend. None of them survive due diligence.
The drafting mistake that kills most provisionals
This is where the failure rate gets ugly. The official conversion rate from provisional to utility in the US is somewhere around forty percent. The rate of provisionals that convert into granted, defensible utility patents is much lower. The reason, almost every time, is the drafting.
Founders treat the provisional as a low-stakes filing because it is cheap. So they upload a slide deck, a few diagrams, maybe a paragraph of description. The USPTO accepts it because the bar for a provisional is low. But when twelve months later they try to convert it into a utility filing, the priority date only protects what was actually disclosed in the provisional. If your provisional was a five-slide pitch deck, your priority date is a five-slide pitch deck.
I have seen founders lose entire claims at conversion because the provisional did not describe the technical mechanism in enough detail. The new utility application introduces specificity, the examiner asks where this came from, and the answer is: it was not in the provisional. That means no priority date for those claims. Which means anyone who filed in the intervening twelve months can leapfrog you.
The fix is not expensive. It is just disciplined. Treat the provisional like a draft utility application, not a placeholder. Write out the technical mechanism. Include the variations and edge cases you might want to claim later. Add diagrams. Spend the money on a competent patent attorney for the drafting, not on filing fees. The filing is the cheap part. The drafting is what determines whether you own anything in twelve months.
The conversion decision is the real test
Twelve months pass quickly when you are running a company. The provisional expires whether you are ready or not. So the conversion decision — whether to file the utility, abandon the provisional, or refile a new provisional — is where the strategy actually happens.
I run founders through three questions when conversion comes due. Is the invention still core to where the company is going? Has the market validated that customers care about this specific capability? And is there a credible scenario where this patent generates real value — defensive cover, licensing revenue, or strategic optionality at exit?
If the answer to all three is yes, convert. The cost of a US utility application — somewhere between £8,000 and £15,000 with attorney fees — is small compared to what a granted patent does for an enterprise sales conversation, a Series A defensibility narrative, or an acquisition diligence pack.
If the answer is no, abandon. There is no shame in letting a provisional lapse. It saves you the conversion cost and the multi-year prosecution that follows. The waste was the provisional itself, not the decision to stop spending more money on it.
The third option — refiling a new provisional on the same subject matter — only works if the invention has materially changed. You cannot just buy yourself another twelve months on the same disclosure. The priority date does not reset.
What I tell founders before they file
When a founder asks me whether they should file a provisional, I ask them four things first. What is the disclosure event you are protecting against. Who is the credible competitor. How specific can you be about the technical mechanism right now. And what does the path to a granted patent look like over the next two years.
If they cannot answer those, the provisional is premature. We talk about whether trade-secret protection is a better fit, whether a defensive publication makes more sense, or whether they are simply not ready to file anything yet. That last conversation saves more money than any drafting tip I have ever shared.
The provisional is a tool. It works when the conditions are right and the drafting is serious. It fails when founders use it as a substitute for thinking. Most of the £2,300 provisionals I see in diligence are the second category. They are not insurance. They are evidence that the founder did not understand what they were buying.
The questions I get asked most
The first one is always: how cheap can I do this? My answer is that the filing fee is not the spend that matters. The drafting is what determines whether you own anything later. A £400 filing on top of a £4,000 drafting engagement is far better value than a £1,500 self-filed provisional that dies at conversion.
The second is whether you can file a provisional yourself without an attorney. Technically yes. In practice, no. The provisional locks in what your priority date covers, and that is not a place to learn on the job.
The third is what happens if you file a provisional, do not convert, and then someone else files the same idea. Once the provisional expires unconverted and your invention has been publicly disclosed, you cannot refile and claim the original date. The window is gone. That is why the conversion decision matters more than the filing decision.
If you want to talk through whether your specific situation justifies a provisional, that is what we do at Beyond Elevation. I run a patentability audit that covers exactly this — what to file, what not to file, and what your path to a granted patent actually looks like.
By Hayat Amin — founder, Beyond Elevation. I advise pre-seed through Series B founders on patent strategy, IP monetisation, and turning research into defensible value.