Which of our products are approved, in which markets, under which applications, since when?
The commercial organisation asks constantly. The supply chain depends on the answer. Regulatory affairs is accountable for it. And in most companies it is answered by a spreadsheet, or by a registration tracker somebody maintains by typing.
Every one of those facts arrived in a letter that is already filed.
The gap between the tracker and the evidence
Most platforms handle registrations properly — the marketing authorisation in force per market, with its application, approval date and status, linked to products and applications. The design is rarely the problem.
The problem is that the only way to create one is to type it. And typing into a tracker that nothing else uses yet is nobody’s priority, so it sits empty while the agency letters that granted those authorisations are already filed, categorised and linked to applications in the correspondence register.
The information is there. Reading it back out is what is missing.
Propose first, then apply
DnXT splits this into two steps, and the split is the design.
The first shows what the confirmed letters would produce, and changes nothing. The second writes the proposals you selected, re-checking first — so what gets written reflects the information as it stands at that moment, not as it stood when the screen was drawn.
Software that changes records as a side effect of somebody opening a report is software people stop trusting. Once is enough.
The rules that stop a suspicion becoming a fact
The correspondence register categorises letters, and some of that is worked out automatically while some is confirmed by a person. Only the confirmed ones count here.
A letter the system believes is an approval letter, but which nobody has confirmed, is reported as unusable with the reason — and is not used. It would be easy to accept it: the categorisation is good, the letter almost certainly is what it looks like, and the resulting entry would be right most of the time.
But a marketing authorisation derived from an unconfirmed guess is a regulatory fact resting on a guess, and nothing downstream can tell it apart from one derived from a confirmed letter. The discipline throughout DnXT’s correspondence handling is that a suspicion never becomes a fact on its own. A person promotes it, or it stays a suspicion.
Beyond confirmation, the letter must be linked to an application and carry a date. Without an application there is nothing to register against; without a date there is no approval date, and a registration with a blank approval date is not usable.
Supplements are not new registrations
The rule that takes the most thought: one active registration per product, market and application.
A later approval letter on the same application is a supplement approval. It is a real regulatory event, and it is not a second marketing authorisation. So it does not create a new registration, and the earliest letter sets the approval date — because the authorisation has been in force since the original approval, not since the most recent supplement.
Getting this backwards gives you a tracker where every product appears approved far more recently than it was: wrong in a way that is hard to spot and awkward to explain to an auditor.
Existing entries are treated carefully too. A pending registration with no approval date gets completed — that is exactly the case this exists to fix. An already-active one is left alone. Overwriting something a person curated with something software worked out is what makes people switch automation off.
Rejections are outcomes, not silence
Complete response letters and refuse-to-file notices grant nothing. They are also among the most consequential letters a regulatory team receives, and leaving them out because they produce no registration would be a strange kind of optimism.
They are shown as outcomes: this letter was found, this is what it was, it grants no authorisation. Alongside them, every letter that could not be used comes back with its reason — not linked to an application, no date, category never confirmed.
That list is quietly the most useful part. It is a work queue for the correspondence register, produced as a by-product. “Twelve letters need linking to an application” is a concrete afternoon’s work. “The registration data is incomplete” is not.
Every entry names its source
A registration produced this way carries a pointer to the document it came from.
That is what makes the whole thing defensible. Ask “why does the system say we are approved in this market?” and the answer is a specific approval letter, in the register, with its own version history and audit trail — not “the software worked it out”, and not “somebody entered it in 2023”.
These entries are stored exactly like manually entered ones, differing only in that they can point at their evidence.
The pattern behind three capabilities
This is the third capability built on the same observation: the fact you are about to ask someone to type is usually already in a document you filed or received.
Facility details are on the Form 356h you submitted. What is in force is in the submissions you published. Marketing authorisations are in the letters the agency sent you.
Reading them back does not just save the typing. It produces a record that points at its evidence — which a hand-maintained tracker can never do.
DnXT builds eCTD publishing, submission planning, document management and dossier review software for regulatory operations teams. Book a demo to see registrations read from the letters that granted them against your own submissions.
