What Companies Look For in an Outside Invention Submission

A submission does not get read the way its author imagines it being read. The inventor pictures a meeting: a room, a discussion, questions about the mechanism. What actually happens is a sort. Somebody with responsibility for a product line and a revenue number opens the file, spends between four and nine minutes on it, and assigns it to one of three piles. The middle pile — worth a second look — is the only one that matters, and it is usually a single-digit percentage of what arrives.

That sorting is not hostile and it is not lazy. A category manager carrying six or eight active lines may see several hundred outside items in a year on top of everything the internal development team is already pushing. The screening exists because the volume demands it. What makes the process feel arbitrary from the outside is that the criteria are never published, rarely explained in a rejection, and almost never the ones the submission was written to satisfy. The document argues that the idea is clever. The screen is asking whether the company can sell it, at what margin, into which slot, and with what exposure.

The Sort · Minutes One to Nine

What Actually Happens in the First Sitting

Most established manufacturers run a formal intake process, and the first document in it is not about the invention at all. It is a submission agreement: a short contract stating that the company accepts no obligation of confidence, that it may already be working on something similar, that it retains the right to develop competing products independently, and that any compensation flows only from a separately negotiated licence. Signing it feels like surrendering something. Refusing to sign it ends the conversation, because no legal department will let a product team look at outside material without it.

The practical consequence is that nothing in a first submission should be secret. Anything genuinely confidential — an unfiled process, an undisclosed material specification, the tolerance that makes the mechanism work — belongs in a filing before it belongs in a package. What goes into the package is the part that can survive being read by somebody under no duty to protect it: the problem, the result, the visible mechanism, the rights position, the numbers.

Screeners work through a submission in a predictable order, and they stop at the first hard failure. Rights, then cost, then fit, then evidence. An idea that fails the rights test is not evaluated on cost. This is why detailed engineering appendices so often go unread — the reader never reached them.

Test One · Freedom

Can We Sell This Without Being Sued

The first question is never "is this patentable". It is "if we tool this, launch it, and put it on shelf in eighteen months, who can stop us or take a cut". Those are different questions with different answers, and conflating them is the most common technical error in a submission.

Patentability asks whether your claims are novel and non-obvious over the prior art. Freedom to operate asks whether making and selling the product would infringe somebody else's live claims. An invention can be perfectly patentable and completely unsellable — the classic case being an improvement to a mechanism that somebody else still holds a dominant patent on. Your narrow claim is valid; practising it still requires their licence. Buyers have been caught by this before and now ask early.

What they want to see is specific and unglamorous. Application or grant numbers rather than the phrase "patent pending". Filing dates and priority dates. The independent claim, quoted, so an in-house attorney can read scope in thirty seconds rather than requesting the file. Confirmation that assignment is clean — that every named inventor has assigned their interest, that no former employer has a claim through an employment agreement, that no co-developer contributed a feature and was never documented. A co-inventorship dispute discovered at diligence stage kills a deal faster than a bad cost model.

Timing matters here too. A provisional application fixes a date and buys twelve months, but it is never examined and never granted; a company reading one is being asked to value an option rather than an asset. A non-provisional publishes at eighteen months from priority and typically draws a first examination report between eighteen and thirty months after filing, which means a submission made in year one is usually made against an unexamined, unpublished record. None of that is disqualifying — most licences are signed on pending applications — but it changes what the company is willing to commit. Getting the sequence and the scope right before anyone else reads the file is precisely where help with patenting invention ideas earns its cost, because the position described in the submission is the position you are stuck with.

Test Two · Margin

The Landed-Cost Arithmetic Nobody Sends

This is where the majority of otherwise good submissions die, and they die silently, because the arithmetic is done by the reader in their head and never communicated back.

Work it backwards from the shelf. Suppose the product plausibly retails at a common price point. The retailer typically takes something close to a fifty per cent margin on that figure, so the company's wholesale price is roughly half of retail. Out of that wholesale figure the manufacturer must cover unit cost, freight, duty, warehousing, returns, marketing support, overhead, its own margin — and the inventor's royalty. Rules of thumb vary by category, but many consumer-goods teams look for a factory cost around a quarter to a fifth of the intended retail price before they will take a concept seriously. Miss that ratio and no amount of cleverness rescues it.

Royalties are the smallest line in that stack and the one inventors expect to be largest. Typical rates for licensed consumer products fall in the low single digits — often two to six per cent of net wholesale receipts, not of retail, and not of gross. On a product wholesaling at a modest figure and selling in five-figure annual units, that is real money and not life-changing money. Presenting a demand for ten per cent of retail signals unfamiliarity with the entire structure.

Then there is tooling. A single-cavity injection mould for a modest part is a five-figure commitment; a multi-cavity family tool for a product with several components climbs quickly. That capital has to be amortised over forecast volume, which is why a design requiring four moulds instead of one is not marginally more expensive but categorically less attractive. The geometry of the part determines this, and the decisions are made early and cheaply — the reason manufacturability deserves attention long before a buyer sees anything. Minimum order quantities compound it: a factory quoting five or ten thousand units as a first run has just set the company's inventory risk, and that number appears in the buyer's calculation whether or not the submission mentions it.

Test Three · Fit

Shelf Slots, Line Reviews and the Reset Calendar

The third test is the one inventors find hardest to accept, because it has nothing to do with the quality of the invention. A new product does not join a range; it displaces something. Shelf space is finite, planograms are set, and adding a facing means removing a facing that is currently generating revenue.

So the reader is asking a comparative question: does this earn more per linear unit of shelf than whatever it pushes out, and does it bring genuinely new volume rather than cannibalising the company's existing lines? An item that is excellent but overlaps an existing product at a similar price is often rejected for exactly that reason. An item that is merely good but opens an adjacent use case can win.

Calendars are just as unforgiving. Major retail categories are reviewed on a cycle — commonly once or twice a year — and a product that misses a line review waits for the next one regardless of merit. Between acceptance and shelf presence, eight to fourteen months is a normal span once tooling, first-article inspection, safety testing, packaging artwork and distribution are sequenced. Submissions arriving three weeks before a review are read, if at all, as material for the following cycle. Understanding when the window opens is a large part of how to approach the people who actually make these decisions.

Test Four · Proof

What Counts as Evidence, and What to Send

By the fourth test the reader is broadly persuaded and now wants to reduce risk. Evidence, in descending order of weight: units sold at a real price to strangers; a functioning prototype that a third party has handled without instruction; independent test data; a costed quotation from a named factory; and only then drawings, renderings and enthusiasm.

Prototypes divide into two kinds and buyers use the distinction constantly. A works-like model proves the mechanism and may look nothing like the product. A looks-like model proves the form and may contain no working parts. One of each is worth more than a single expensive hybrid, and a photograph of a works-like model in a hand, mid-use, is worth more than either. Where a product touches a regulated area — anything electrical, anything used by children, anything contacting food or skin — the reader will silently price the applicable safety standard and third-party lab testing into the launch cost, and will notice whether the submission shows awareness that this expense exists.

The package itself should be short. One page that a buyer can absorb standing up, backed by material available on request: that is the format that gets read, and the discipline of compressing a product onto a single page tends to expose whichever of the four tests has not yet been answered. Accounts of how individual products actually made the journey are useful calibration here — the published record of how the MixAid moved from a household frustration to a manufactured product shows the sequence at normal speed, and a closer account of the same submission and the people behind it fills in the parts that summaries skip.

Read in that order, the four tests explain most rejections that arrive without explanation. A submission that leads with the mechanism and never mentions cost has answered the least important question at length. One that quotes a claim, states a factory cost, names the slot it intends to occupy and shows a prototype in use has answered all four in a page, and will be read by somebody looking for a reason to say yes rather than a reason to stop. The general sequence — record, protect, cost, present — is set out plainly in this guide to moving an invention idea toward a finished product, and there is nothing in it that requires a large budget, only order.

The company is not evaluating your invention. It is evaluating a decision it has to defend internally — and the submission that wins is the one that makes that defence easy to write.

Rights · Cost · Slot · Proof