The Most Expensive Sentence in Your Submission
Same device, same data, same algorithm. Change a few words about who it's for and what it concludes, and you can cross a cost boundary worth millions.
Most founders assume the FDA pathway is a property of the device. Build something novel, and you're headed for De Novo. Build something conventional, and you're headed for a 510(k).
That isn't quite how it works. The pathway is determined by how you describe what the device does — the intended use statement and the indications that follow from it. Same hardware, same algorithm, same clinical data. Change a few words about who the device is for and what conclusion it supports, and you can move across a cost boundary worth millions of dollars. We see this often enough that we now treat indication language as a budget decision rather than a labeling exercise. Five recent engagements make the point.
What a sentence costs
A company developing a flow cytometry reagent kit for an autoimmune kidney condition came to us with claim language that read as a standalone diagnostic. That framing pointed toward a Class III burden. Narrowing the intended use to an adjunctive role kept the program on a 510(k) route and preserved a defensible predicate argument. The revision saved an estimated six months against a De Novo timeline and avoided rework we valued at more than $500,000.
A digital health company with an AI-enabled sleep apnea application had drafted an indication that positioned the software as a diagnostic tool. Reframed as an adjunctive pre-screening aid, the same product supported a substantial equivalence argument instead of triggering a De Novo assessment. Estimated avoided spend: $2 million to $4.5 million in clinical and operational costs.
A neurostimulation company with an existing 510(k) clearance wanted to pursue a new neurological indication. The language under consideration implied the device changed the course of the condition. Language describing symptom reduction supported a De Novo route. The difference between those two sentences was the difference between a De Novo and a PMA, and we estimated the cost avoidance at $2.5 million to $4 million.
A pediatric communication software platform turned out to be 510(k)-exempt — genuinely, verifiably exempt, requiring only establishment registration and device listing. But we identified two specific ways the company could void that exemption, the most likely being an expansion of the described use into clinical decision support. The product could launch immediately. A paragraph of marketing copy could have undone that.
A behavioral support platform for patients on GLP-1 medications was operating as wellness software in four countries and preparing to enter the U.S. We evaluated each function separately against FDA's general wellness policy and its guidance on multiple function devices. Some features could launch as-is. Others would move the platform into Class II. Mapping that line function by function meant the company didn't prepare a submission for functions that never needed one — preparation that routinely runs $50,000 to $250,000 and up before user fees.
Why this keeps happening
The intended use statement usually gets written twice, by two different groups, for two different audiences.
The first version is written for investors and customers. It's ambitious on purpose. It describes what the product will eventually do, because that's what raises money. The second version is written for FDA, and it has to be defensible against the evidence that actually exists today.
The trouble starts when the first version becomes the second version by default — when the language in the pitch deck migrates into the regulatory documentation because nobody stopped to ask what each phrase obligates the company to prove. Words like "detects," "diagnoses," "predicts," and "identifies patients at risk" each carry a different evidentiary burden. So does the distinction between a tool a clinician consults and a tool a clinician relies on.
None of these are wrong words. They're expensive words, and they should be chosen deliberately.
Settle the sentence before you spend the slot
A Pre-Submission is designed to test a pathway, not to discover one. If your indication language is still unsettled when you request the meeting, you'll spend a limited FDA interaction on a question you could have answered internally — and you may receive feedback shaped around a version of the product you're about to change.
Before the Pre-Sub request goes out, three things are worth resolving:
Write the indication in its narrowest defensible form first. Start from what your current evidence supports, not from what the roadmap promises. You can always plan an expansion. It's far harder to walk a claim back after FDA has responded to it.
Test the variants against their consequences. Draft two or three versions of the indication and map each to its likely classification, evidence requirements, and timeline. The exercise takes days. The gap between the versions is frequently measured in years.
Separate the launch claim from the growth claim. Several of the engagements above resolved into a phased strategy: a narrow initial authorization that generates revenue and clinical experience, followed by a planned expansion supported by data the first phase produces. That sequencing is a regulatory decision and a commercial one at the same time.
The companies that handle this well aren't more cautious than their peers. They're more specific, earlier — and specificity is what keeps the FDA conversation focused on the product you're actually bringing to market.
If you're drafting or revising an indication statement ahead of an FDA interaction, a scoped regulatory memo resolves the pathway question before you commit the budget.
Related reading: The Strategic Second Opinion: Why One FDA Pre-Submission Is Rarely Enough