How we decide what not to patent

Filing everything is how you waste a patent budget. Deciding what to set aside, and why, is where a lot of the value actually is.

Here is something you might not expect a patent attorney to say: part of the job is talking clients out of filing.

It sounds backwards. Isn't a patent practitioner supposed to want you to file, and file often? But filing everything is precisely how a company wastes its patent budget. The real skill, the one that actually protects a business, is judgment about what deserves protection and what doesn't. And sometimes the most valuable advice is: not this one, or not this way.

Deciding what not to patent is not a failure of the process. It is the process. Here is how we think about it.

1. Can you tell if someone infringes it?

A patent is only as valuable as your practical ability to detect and prove infringement. A granted patent gives you the right to stop others, but exercising that right requires knowing they're doing it and being able to show it.

So the first question for any candidate is: if a competitor used this, would you be able to tell? Some inventions are visible in a shipped product, a user interface, a device you can buy and inspect, a protocol on the wire. Those are detectable. Others live somewhere you will never see: deep inside a competitor's server, buried in a training pipeline, embedded in an internal manufacturing step. If infringement happens entirely behind a competitor's walls, a patent on it can be extraordinarily difficult to enforce, and in exchange for that hard-to-enforce right, you will have published a detailed roadmap of exactly how you did it. Low detectability is one of the strongest reasons to pause before filing.

2. Would a patent teach competitors more than it protects?

A patent is a public disclosure. That is the fundamental bargain: in exchange for a time-limited right to exclude, you tell the world exactly how your invention works, in enough detail that a skilled person could reproduce it. For most inventions, that trade is worth it. For some, it isn't.

This is the trade secret fork. If an invention can be kept genuinely confidential, and is hard to reverse-engineer from your product, a trade secret may protect it far longer than a patent would. A patent expires in twenty years; a trade secret can last indefinitely, as long as it stays secret. Manufacturing processes, certain formulations, some backend methods, and internal techniques a customer never sees can all be candidates. The classic example is Coca-Cola, whose formula has been protected as a trade secret for over a century rather than patented, because a patent would have disclosed it and then expired, handing it to everyone.

The fork is genuinely a fork, though, not a default. Trade secrets carry their own risks: independent discovery is a complete defense, reverse engineering is generally lawful, and once the secret escapes, it's gone. If an invention is likely to be independently developed or reverse-engineered, a patent's public right may be the stronger protection despite the disclosure. The point is that it's a real decision to make deliberately, per invention, not a reflex to patent everything.

3. Will it still matter by the time it grants?

Patents take years to issue. That timeline matters when you're deciding what to file. If an invention is likely to be obsolete, superseded, or simply irrelevant to your product roadmap within that window, the money and attention spent prosecuting it may never earn a return.

This calls for judgment, not a blanket rule. Fast-moving implementation details, the specific way you built something this quarter, sometimes fall into this bucket, while the underlying architecture or approach beneath them may be exactly what's worth protecting for the long term. Separating the durable core from the disposable surface is part of deciding what to file.

4. Is it core, or is it noise?

Engineering teams generate clever solutions constantly. Not all of them are strategically important, and treating every clever thing as a filing candidate is a fast way to exhaust a budget on inventions that don't move the needle.

The question is not "is this patentable." Plenty of things are patentable and still not worth patenting. The question is "does protecting this advance the business", does it create real defensibility, block a competitor's likely path, or add value an investor or acquirer would recognize? Filing on peripheral inventions drains budget and focus away from the handful that actually anchor a portfolio.

Why this matters more under a budget

For a large company with a deep IP budget, over-filing is inefficient. For a growing company with a finite budget, it's genuinely dangerous, because every filing that shouldn't have happened is a filing that should have, and didn't. The inventions that anchor a portfolio, the ones a competitor can't avoid and an acquirer pays attention to, deserve the budget. Getting there requires the discipline to set the others aside.

That's why an honest evaluation process is as much about exclusion as inclusion. When we look at a set of invention candidates, we're asking, for each one, not just "can this be patented" but "should it be, and if so, is a patent even the right tool." The filtering is not overhead on top of the real work. It is the real work.

The strategy changes the calculus

Everything above describes a fairly common situation: a company deciding, with a finite budget, which inventions to protect around its own products. But the right approach genuinely depends on what you're trying to accomplish, and some strategies shift the calculus in important ways.

A company deliberately building a licensing or assertion portfolio weighs these questions differently. When the portfolio itself is the asset, rather than a shield around a single product, breadth and coverage can matter more than they would for a purely defensive filer. Some inventions worth setting aside under a lean, product-focused strategy are worth protecting under a portfolio-building one, because their value lies in strengthening the estate's overall coverage, not in reading on your own roadmap. Detectability still matters (you have to be able to identify use to license or assert), but the threshold for "worth filing" moves.

Timing is a lever, too. The obsolescence concern above is partly a function of how long prosecution takes, and that isn't fixed. Accelerated examination options can compress what would normally be years of prosecution into a matter of months. Where speed matters, an invention in a fast-moving space, or a patent you need granted quickly to support assertion, a licensing conversation, a financing, or a transaction, fast-tracking can change the math on whether something is worth filing at all. An invention that looked too fast-moving to protect under a normal timeline may be very much worth protecting on an accelerated one.

So the questions above are a starting point for the conversation, not a fixed rule. The right answer depends on the invention, the business, the competitive landscape, and the goals behind the portfolio.

The mindset

These are all case-by-case calls. Detectability, the trade secret fork, obsolescence, strategic weight, every one depends on the specific invention, the specific business, and the specific competitive landscape, and none of them reduces to a rule you can apply blindly. But the mindset behind them is constant: the goal was never to file the most patents. It's to own the right ones. And owning the right ones starts with the discipline to say no to the rest.