ResultsClock

1. One-liner

Warns university research offices which trials hit the ClinicalTrials.gov results deadline before the $10,000-a-day meter starts.

2. Trend signal — why now?

On 30 March 2026 the FDA sent targeted reminder messages to more than 2,200 sponsors and researchers covering more than 3,000 registered clinical trials that had no results posted. It disclosed this publicly on 13 April 2026, stating that 29.6% of studies highly likely to fall under mandatory reporting requirements have no results information submitted. The agency described the messages as “an extra step” before considering “further regulatory action” — and confirmed it may issue Pre-Notices of Noncompliance and Notices of Noncompliance.

That is a regulator announcing, in writing, that the informal-tolerance era is over.

The money is not theoretical. Under Section 303(f)(3)(B) of the FD&C Act, a violation not corrected within 30 days of notification accrues an additional penalty per day, per trial — Mount Sinai’s own research-administration blog puts the current figure at $13,237 per day of non-compliance and notes more than 60 Pre-Notices have been sent to date. Worse for the buyer: under Section 402(j)(5)(A) of the PHS Act, NIH can withhold remaining or future grant funds, and where a noncompliant trial meets the applicable-clinical-trial definition, NIH can suspend funding to the recipient on an institution-wide basis.

The gap is structurally academic, not universal. At five years after trial completion, only 27.7% of trials funded by academic institutions or other government sources had posted results, versus 41.5% for industry-funded trials. Industry sponsors have regulatory-affairs departments; academic investigators frequently do not realise results entry is a separate, substantial data-submission task from registration.

I pulled the live numbers myself rather than trusting a vendor blog. Querying the public ClinicalTrials.gov API v2 for trials with a primary completion date between 1 Sep 2025 and 31 Aug 2026 — i.e. those whose 12-month results deadline lands in the next twelve months:

Lead sponsor class	Trials
OTHER (universities, hospitals, foundations)	9,406
INDUSTRY	2,479
NIH	70

Non-industry sponsors own roughly four times the deadline volume of industry. In a 1,000-trial sample of that OTHER cohort there were 638 distinct sponsors, only 168 of which had two or more trials — a long tail of institutions running one or two trials each, with no full-time disclosure staff and no reason to buy enterprise software.

Provenance:

• Signal 1 (demand): FDA reminded >2,200 sponsors across >3,000 trials on 30 Mar 2026; 29.6% of likely-applicable studies have no results posted — https://www.fda.gov/news-events/press-announcements/fda-reminds-more-2200-sponsors-and-researchers-disclose-trial-results — observed 2026-09-06
• Signal 2 (feasibility): ClinicalTrials.gov API v2 is a free, public, no-authentication REST/JSON API exposing primary completion date, results-posted status, sponsor and sponsor class — https://clinicaltrials.gov/data-api/api — verified live 2026-09-06 (returned 9,406 OTHER-class trials for the Sep 2025–Aug 2026 window)
• Signal 3 (economic): Penalties run to $13,237/day per uncorrected trial plus institution-wide NIH funding suspension; Mount Sinai policy makes departments financially responsible for their faculty’s penalties — https://researchroadmap.mssm.edu/blog/spotlight-clinicaltrials-gov-registration-and-reporting-enforcement/ — observed 2026-09-06
• Signal 4 (incumbent shape): FDAAA TrialsTracker (Oxford Bennett Institute) publicly lists overdue trials but explicitly does not help sponsors submit results and does not send advance alerts before deadlines — https://fdaaa.trialstracker.net/faq/ — observed 2026-09-06 Category: Workflow automation (a dated, penalty-backed deadline whose detection layer is a free public API and whose owner is an under-staffed research office)

3. The opportunity

There is a free, well-built, Oxford-maintained tool that tells the world which trials are overdue. It is called the FDAAA TrialsTracker and it updates every working day. It is also, from the buyer’s point of view, precisely the wrong product — and its own FAQ says so.

The TrialsTracker does not help sponsors submit results and does not send advance alerts before deadlines; it reports trials after they become overdue. It is a public accountability board. By the time your institution appears on it, the 12-month clock has already expired, the trial is publicly flagged as “Late Results,” and you are in the window where a Notice of Noncompliance starts a 30-day fuse on a five-figure daily penalty.

That is the entire opening. Nobody owns the interval between primary completion date and deadline. The regulator publishes the trigger. Oxford publishes the failure. The eleven months in between — where the work actually is, and where the penalty is still avoidable — has no product pointed at it.

The enterprise vendors that exist (TrialAssure REGISTRY, Citeline’s TrialScope Disclose) are real, capable, multi-registry disclosure suites built for pharma sponsors filing across 30–40 global registries. Neither publishes pricing; both route you to a demo and a specialist. They are sold to organisations that already employ disclosure managers. A hospital research office with nine open trials and one PRS administrator who also runs IRB submissions is not buying a global disclosure suite, and those vendors are not structurally able to serve them at that price point — TrialAssure has publicly acknowledged the mismatch by creating a separate higher-education discount programme.

So the market splits cleanly: pharma buys the suite, and 9,406 trials a year sit with sponsors who buy nothing and find out they had a problem when Oxford publishes it or the FDA emails them.

The 10× is not in the data — the data is free. It is in converting a date into a work queue with enough lead time to act on it, and in doing the one thing the free tracker refuses to do: help you actually get the results submitted.

4. Target market

• Primary customer: The ClinicalTrials.gov PRS administrator or research-compliance manager at a US academic medical centre, teaching hospital, or hospital-based research institute — the person named in institutional policy as responsible for releasing records on behalf of the university. Typically one to three people covering 20–300 registered studies, sitting inside an Office of Research / Clinical Trials Office. Real examples visible in the live data: University of Michigan, Emory, Yale, Mayo Clinic, Brigham and Women’s, Massachusetts General, University of Wisconsin–Madison.
• Secondary customer: The department administrator at institutions like Mount Sinai, whose policy “makes departments responsible for any financial penalty incurred by their faculty member due to non-compliance.” When the fine lands on a department’s budget rather than the central office’s, the department will buy its own early-warning system.
• Why they buy: Because the failure mode is career- and budget-damaging and entirely preventable. UNC’s own guidance states plainly: “For NIH-funded studies, failure to comply may lead to NIH withholding future grant funding for the institution.” One forgotten trial from a PI who has since left can jeopardise institution-wide funding. The current control is a spreadsheet and the PRS administrator’s memory.
• Rough TAM reasoning: 9,406 non-industry trials per year hit the deadline, spread across a long tail where a 1,000-trial sample held 638 distinct sponsors. Restricting to US institutions with recurring trial volume (2+ trials), the serviceable base is realistically 800–1,500 institutions and departments in the US, plus a meaningful spillover of non-US academic sponsors (Cairo University alone had 23 trials in the sample) who register on ClinicalTrials.gov and face the same posting rules for applicable trials. At $300–800/mo that is a $4–14M ceiling — comfortably a $1–5M business without needing to win the whole market.
• Why now for them: March 2026. The FDA moved from silence to 2,200 targeted messages and told everyone the next step is Notices of Noncompliance. FDA even ran a dedicated session, “ClinicalTrials.gov: Essentials for Academic Medical Centers,” on 14 July 2026. Research offices that deferred this for a decade now have a documented reason to spend money on it this budget cycle.

5. Product sketch (MVP)

• Institution roster, auto-built. Enter your organisation name; the product pulls every trial where you are lead sponsor from the public API and builds the portfolio without any manual entry or PRS credentials.
• The countdown. Every trial gets a single number: days until its results-submission deadline, computed from primary completion date. Sorted by urgency, colour-coded at 180 / 90 / 30 days, and — critically — surfaced before the deadline, not after.
• Applicability triage. Flags which trials are likely “applicable clinical trials” subject to mandatory reporting and which probably are not, with the reasoning shown. This is the judgment call the free tracker admits it cannot always make, since ClinicalTrials.gov does not publish ACT status.
• Orphan-trial detection. Finds trials whose PI has left, whose record has not been updated in 24+ months, or whose status is stale — the ones nobody remembers owning. These are the trials that generate the penalties.
• Results-readiness pack. For each approaching trial, a checklist of the four results modules required (participant flow, baseline characteristics, outcome measures, adverse events), what data the team must assemble, and a drafted submission-ready summary from the protocol record for the PI to correct rather than compose.
• QC-comment coach. Paste the PRS reviewer’s comments; get a plain-English explanation of each Major Comment and what change resolves it. QC review runs under 30 days but the resubmit cycle is the step that pushes teams past their deadline.
• Escalation emails. Automatic notices to PI, department administrator and research office at fixed thresholds, with the running penalty exposure stated in dollars. The dollar figure is what unsticks a non-responsive PI.
• Board-ready exposure report. One page: trials at risk, days remaining, aggregate daily penalty exposure if all lapse, and institution-wide NIH funding risk. This is the artefact that justifies the subscription to whoever signs.

6. AI angle — what’s load-bearing

Strip the AI out and roughly half this product survives — the countdown itself is arithmetic on a public API, and I would rather say that plainly than pretend a date subtraction is machine learning.

Where AI genuinely carries weight:

Applicability triage. Deciding whether a given trial is an “applicable clinical trial” is a legal-definitional judgment over unstructured protocol text — intervention type, phase, study design, US site presence, FDA-regulated product. ClinicalTrials.gov does not publish ACT status, and the Oxford tracker states it infers this and cannot always be certain. A model reading the full protocol record and producing a reasoned, cited classification is doing work that otherwise requires a compliance officer’s afternoon, per trial, across hundreds of trials.

Drafting the results narrative. The outcome-measure and adverse-event modules are where submissions stall and where QC comments land. Generating a structured first draft from the registered protocol — correct arm labels, correct outcome definitions, correct units — converts a blank-form problem into an editing problem. That is the difference between a PI doing it this week and deferring it another quarter.

Decoding QC comments. PRS reviewer comments are terse and reference internal review criteria. Translating “Major Comment” text into the specific field-level fix is exactly the pattern-matching an LLM does well and a first-time submitter does badly.

The honest framing: AI is not what makes the product exist, it is what makes the product finish the job instead of just ringing an alarm. And finishing the job is the whole differentiator against a free tracker that only rings alarms — late.

7. Localization angle (if any)

N/A — this is a US-first regulatory play. The penalty structure (FD&C Act §303(f)(3)(B)) and the funding lever (PHS Act §402(j)(5)(A), NIH institution-wide suspension) are specifically American, and those two levers are what make anyone pay. Non-US academic sponsors registering on ClinicalTrials.gov are addressable spillover, not a localisation strategy.

The genuine second market is not a country, it is a registry: the EU’s CTIS carries a parallel results-disclosure duty, and the same countdown-and-draft mechanic ports directly. That is the v2 expansion, not a v1 wedge — and it is also where the enterprise incumbents are strongest, so I would not lead with it.

8. Business model — path to $1M–$5M ARR

• Pricing: Tiered by portfolio size, sold to the office not the seat.
  • Department — $249/mo, up to 15 trials
  • Research office — $599/mo, up to 75 trials
  • Institution — $1,200/mo, unlimited trials, multi-department roles, board report
• ACV: ~$7,200 blended. Most early revenue comes from the $599 tier — the single research office with a few dozen studies is the sweet spot.
• Rough math to $1M ARR: 140 customers at $599/mo = $1.006M. That is roughly 10–17% of the realistic US serviceable base — achievable without winning the flagship institutions.
• Rough math to $5M ARR: ~580 customers at blended $7.2K, which requires moving upmarket into institution-tier deals at the large academic medical centres and adding CTIS/EU registry coverage to lift ACV toward $12–15K. Alternatively 350 institution-tier accounts. This is the tier where you start colliding with TrialScope and TrialAssure, so $5M is a genuine fight; $1–2M is not.
• Expansion path: Trial-count growth is automatic and non-negotiable (institutions register more studies every year, and each one eventually hits a deadline). Then: EU CTIS coverage, then the managed-service upsell — “we draft and submit the results for you” at $2–4K per trial, which is where the margin actually lives and where under-staffed offices will happily convert capex into someone else’s problem.

9. Go-to-market wedge — first 100 customers

The distribution advantage here is unusual and worth being explicit about: the prospect list is a public API query, and it includes the severity of each prospect’s problem.

• Run the query, send the evidence. Pull every US non-industry sponsor with trials whose deadline falls in the next 180 days and no results posted. That is a named institution, a named trial NCT number, and a specific date. Cold email the research office: “Your trial NCT0XXXXXXX has a results deadline on [date]. Here are the four other trials in your portfolio in the same window.” Not a pitch — a free audit of their own public exposure. This converts because it is verifiable in ten seconds and because it is genuinely their problem.
• Harvest the FDA’s own enforcement wave. The FDA contacted 2,200+ sponsors in March 2026 across 3,000+ trials. Overdue trials are publicly identifiable via the TrialsTracker and the API. Those institutions have an internal, active, budget-approved fire drill right now. Lead with remediation (“here is what it takes to clear these”), not prevention.
• Sell to the department, not the university. Mount Sinai-style policies push the financial penalty onto the department that employs the faculty member. Department administrators have discretionary budget, no procurement committee, and direct exposure. A $249/mo departmental tier closes in one call where an institutional deal takes two quarters. Land there, expand to the central office.
• Go where the PRS administrators already gather. SOCRA and ACRP chapters, the Association for Clinical and Translational Science, and university research-administration listservs. FDA ran “ClinicalTrials.gov: Essentials for Academic Medical Centers” in July 2026 — the attendee profile for that session is the customer. Present the live compliance-rate data (27.7% academic vs 41.5% industry) as a talk, not a demo; the audit offer is the CTA.
• Publish the league table. An annual, sourced “academic disclosure compliance ranking” built from the public API. Oxford proved this attracts institutional attention; nobody has paired it with a product that fixes the ranking. Being named is the lead-gen mechanism.

10. Build complexity — justification

Low. The data layer is a free public REST/JSON API with no authentication and no rate-limit negotiation — I verified it live and had sponsor-class trial counts and per-institution portfolios back in a single query. There is no scraping, no credential handling, no data-acquisition problem, and no PHI: everything the product reasons over is already public. That removes the two things that normally make health-tech slow — data access and compliance surface.

The custom work is the applicability-classification logic, the results-draft generator, and the notification/escalation engine, plus a straightforward multi-tenant dashboard. A competent solo builder ships a credible v1 in 6–8 weeks; the free-audit lead magnet is a weekend script that can start generating pipeline before the product exists.

The honest caveat: the managed submission service upsell is operations-heavy and needs someone who has actually filed results in PRS. That is a month-6 decision, not a v1 dependency.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Reads only public ClinicalTrials.gov data; no PHI, no credentialed access, no regulated advice being sold.
Ethical — no harm / dark patterns	✅	Increases public disclosure of trial results, which is the explicit policy goal. Penalty figures quoted are the regulator’s own.
Market exists (evidence above)	✅	9,406 non-industry trials hitting deadline in a 12-month window; FDA contacted 2,200+ sponsors March 2026; paid incumbents exist upmarket.
1–5 person team can build this	✅	Public API + LLM + dashboard. 6–8 weeks solo.
Launchable with <$50K / ₹40L	✅	No data costs, no licences, no sales team required to open the funnel. Inference and hosting only.

All five pass.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	17/20	$13,237/day per trial plus institution-wide NIH funding suspension, and departments are made to eat the fine. Not 18+ because the pain is episodic per trial rather than daily, and many sponsors have quietly absorbed the risk for a decade.
Demand evidence	15	13/15	FDA’s own March 2026 action across 2,200+ sponsors, published 29.6% non-posting rate, 27.7% vs 41.5% academic/industry gap, paid incumbents upmarket, and live API counts I ran myself. Short of 15 only because I have no confirmed dollar spend from an academic buyer at this price point yet.
Build feasibility	15	13/15	Free public API, no auth, no PHI, verified working end-to-end. Applicability classification is the only genuinely hard piece.
Distribution clarity	15	11/15	The prospect list and their exposure are both public — outbound writes itself. Docked because research-office procurement is slower than the email response rate suggests, and the true buyer is sometimes ambiguous between department and central office.
Revenue mechanics	15	10/15	Pricing is inferred, not benchmarked — incumbents hide theirs. Academic budgets are real but slow and grant-cycle bound. The $1M path is credible; $5M needs the managed-service or CTIS expansion to work.
Time to first revenue	10	8/10	Free audit → paid conversion is fast because the deadline is a date, but expect 4–8 weeks and a purchase-order process rather than a credit card.
Defensibility	10	6/10	Data is public, so the moat is not data. It is the accumulated applicability-classification logic, the QC-comment corpus, workflow lock-in inside the research office, and brand in a small, tight-knit professional community. A funded competitor could copy the countdown in a month; copying the submission workflow and the trust takes a year.
Total	100	74/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required

The build is light, but selling to research-compliance offices requires credibility in their language. A founder who has worked in clinical research administration — or a co-founder/advisor who has personally submitted results in PRS — is close to mandatory. Without that, the QC-coach and applicability features will be subtly wrong in ways the buyer detects immediately.

Key assumptions to validate (3–5)

1. Assumption: Research offices will pay $599/mo for early warning when a free tool already lists their overdue trials. How to test: Run the free-audit email to 100 institutions with trials due in <180 days; measure how many book a call, then present pricing to 20 and count how many request a quote or PO rather than declining outright.
2. Assumption: The department administrator (not central research office) is the faster-closing buyer where policy pushes penalties onto departments. How to test: Split outbound 50/50 between department administrators and central offices at Mount Sinai-style institutions; compare reply rate and time-to-first-call.
3. Assumption: Applicability classification can be done accurately enough to be trusted. How to test: Classify 200 trials with known ACT status (cross-referenced against TrialsTracker’s inferred set and any FDA notice recipients); target >90% agreement and inspect every disagreement by hand.
4. Assumption: The results-draft generator saves enough time that PIs actually finish submissions. How to test: Run 5 real trials end-to-end with pilot institutions; measure hours from kickoff to PRS submission versus their historical baseline.

Risk flags

1. Incumbent-goes-downmarket risk: TrialAssure already created a higher-education discount programme and offered free academic REGISTRY subscriptions in the past. If they productise a cheap self-serve academic tier, the price wedge narrows fast. The defence is the countdown-plus-draft workflow and speed, not price.
2. Free-tool expansion risk: The Bennett Institute is well-funded, academically motivated, and already owns the data pipeline and the brand. If they add pre-deadline alerting, the top of this funnel evaporates overnight. They have shown no interest in helping sponsors submit — which is the durable half of the product — but this is the single biggest external dependency. Watch their changelog.
3. Enforcement-follow-through risk: FDA has talked tougher than it has acted for a decade; the March 2026 messages were explicitly “reminders,” and only ~60 Pre-Notices have issued to date. If no meaningful penalties actually land in the next 12 months, urgency decays and this becomes a nice-to-have. The NIH funding lever is the more reliable motivator and should be led with.
4. Budget-cycle risk: Academic research offices buy on grant and fiscal cycles, not on urgency. Expect slower conversion than the pain level implies, and price/package for a PO rather than a card.
5. Platform dependency: Everything rests on the ClinicalTrials.gov API v2 remaining free and public. It is government infrastructure with a stated modernisation commitment, so the risk is low — but it is a single point of failure and the Bennett Institute has already been broken once by an upstream change (their pipeline went stale from Feb 2024 until March 2025).

14. Structured verdict

Score:                  74/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical solo founder with a clinical-research-administration
                        advisor who has personally submitted results in PRS
Time to revenue:        6-10 weeks (audit-led outbound; expect PO, not credit card)
Capital to launch:      $5-8K (inference + hosting + a conference booth)
Top 3 assumptions to validate first:
  1. Willingness to pay $599/mo when a free tracker exists — free-audit outreach
     to 100 at-risk institutions, count quote requests from 20 priced conversations
  2. Department vs central-office buyer — split-test outbound, compare time-to-call
  3. Applicability classification accuracy — 200 trials against known ACT status,
     require >90% agreement
Kill criteria:
  - Abandon if <5 of 100 audited institutions book a call within 3 weeks
  - Abandon if applicability classification cannot clear 90% agreement on the
    200-trial benchmark
  - Abandon if the Bennett Institute ships pre-deadline alerting AND a submission
    assistant before v1 launches
  - Abandon if no US institution will commit to a PO within 90 days of first demo

15. Next step — 1-week validation sprint

• Day 1–2: Pull the full at-risk list from the API — every US non-industry sponsor with a trial whose results deadline falls inside 180 days and hasResults: false. Rank by institution. This is simultaneously the market sizing and the prospect list. Hand-verify 20 records against ClinicalTrials.gov to confirm the deadline logic is right before emailing anyone a number about their own institution.
• Day 3–4: Send 100 free-audit emails — 50 to central research offices, 50 to department administrators at institutions with department-liability policies. Each email names their specific trials, the specific dates, and the daily penalty figure with the FDA citation. No product, no pricing, no demo link: just their exposure and an offer to walk through it.
• Day 5: Hold every call that books. In each, ask two questions that produce falsifiable answers: “Who currently tracks these dates, and what tool are they using?” and “If this were $599/mo, who signs — you, or someone else?”

Go/no-go: Proceed only if ≥5 of 100 book a call within the week and ≥2 of those calls name a specific budget holder without hedging. Below that, the pain is real but unowned — and an unowned pain does not convert, no matter how large the fine.