How to decide which inventions are worth patenting
A consistent rubric for scoring candidates, so your filing budget goes where it actually earns its keep.
Once you're surfacing inventions systematically, you face a better problem: you have more candidates than you should file. Filing everything wastes money on patents that never matter; filing nothing wastes the inventions. The value is in the decision, and a good decision needs a consistent rubric rather than gut feel. Here's a framework you can actually use.
Five questions for every candidate
1. Novelty and non-obviousness
Is this genuinely new, and is it more than an obvious combination of what already exists? This is the legal threshold, but it's also a business filter: if a competent engineer in the field would arrive at the same solution, a patent on it will be both hard to get and easy to design around. A quick prior-art look here saves money downstream.
2. Eligibility
Some subject matter, particularly in software, faces higher hurdles to patentability. Abstract ideas implemented on a generic computer are vulnerable; specific technical improvements to how a system works are far stronger. Asking "what concrete technical problem does this solve, and how" early tells you whether a candidate is worth pursuing.
3. Detectability
This is the question startups most often skip, and it's decisive. If a competitor used your invention, could you tell? A patent on something observable in a product, an interface, an output, a measurable behavior, is enforceable. A patent on an internal method no one can see from outside may be impossible to enforce, which makes it a candidate for a trade secret instead. Detectability often matters more than novelty for real-world value.
4. Strategic fit
Does this invention protect something central to your business and roadmap, or is it a clever side-feature you'll abandon in two releases? Filing budget should concentrate on the technology that creates competitive distance in the direction you're actually heading.
5. Continuation runway
Strong inventions often support a family of protection over time, not a single patent. Assessing whether a candidate has room for future claims, as the product and market evolve, helps you decide where to invest in keeping options open versus filing once and moving on.
Run each candidate through those five questions and you get more than a yes/no. You get a routing decision: file a patent, protect it as a trade secret, or hold and revisit. Every candidate lands somewhere deliberate, and your filing budget goes where it earns its keep.
Why a standard rubric matters
Scoring candidates the same way every time does three things. It makes filing decisions defensible, you can show a board or an investor why you filed what you filed. It removes recency and enthusiasm bias, the loudest engineer's pet project doesn't automatically win. And it produces a documented trail, so even in a quiet quarter, you can show that candidates were evaluated and decisions were made on purpose.
Where this connects
Evaluation is the hinge of the whole system: harvesting feeds it candidates, and its output drives what gets filed and how the portfolio is built. It's also, honestly, where a great deal of value is won or lost: the decision of what to protect can matter as much as the drafting of any single filing. The companion articles on harvesting and on the cost of filing round out the picture.
Thinking about your patent strategy?
An intro call covers where your IP stands today, how the practice works, and whether a fractional model fits how you're building. No pitch deck, no obligation.
Book an intro call