Welcome to the first issue of The Essential Record. This is a place for honest, practitioner-level conversations about trial documentation: the TMF, the site file, and the quality of the records they hold. In each issue we take one real question that people in this field actually argue about, and we work through it together, in the words of the people who do the work.
The Essential Record is editorially independent and vendor-neutral. Technologies and platforms may be discussed where relevant, but never as product promotion or paid endorsement. If a topic is yours, we would love to have you in a future issue.
They have partnered with us to put Menka AI into their trial documentation.
The telephone book problem
Why a TMF can be 96% complete and still fail to tell the story of the trial — four people who live inside the file on what to measure instead.
The number
Every good, validated eTMF platform shows you a completeness score. A percentage, a bar, usually green. It is the first thing people ask to see in a demo, and it is the number that ends up on the slide.
A few weeks ago I posted a question about that number on LinkedIn, half expecting silence. Instead, a real conversation broke out underneath it — quality leads, auditors, people who have spent their careers inside the Trial Master File. This piece is my attempt to do that conversation justice, in their words more than mine.
Here is what nagged me enough to post. Completeness compares your TMF against a list of documents you decided, in advance, to expect. That is the whole calculation. A document that never made the list is not a gap. It is not amber. It simply is not there — until an audit finds it for you.
And one of the first replies handed us the image this piece is named after. Barbara Janje-Bode, a TMF expert we will hear more from below, wrote that a TMF can resemble an old telephone book: extensive, meticulously ordered, apparently complete — and still unable to tell you the (hi)story of the city and what actually happened to the people connected to the telephone numbers. We will come back to why.
Because the regulators keep making the same point in drier language. In the EMA GCP inspectors’ 2024 annual report (EMA/INS/GCP/37608/2025), essential documents were the single largest category of major and critical findings. A dashboard can look reassuring — the bar is green — while significant deficiencies remain discoverable in inspection.
So: if your TMF is 96% complete, what does the 96% actually prove?
The shape problem
Eldin’s deeper objection is not to the counting. It is to the shape of the thing being counted. For decades we have sorted documents into zones and folders — a section for ethics, a section for monitoring, a section for safety. But that is not how anyone actually works. When you come back from a monitoring visit, you want everything from that visit in one place: what you planned, what you did, what followed.
The work is process-shaped. The file is folder-shaped. The two never quite meet, and the space between them is exactly where inspectors go looking.
He would go further still: stop forcing every record into a single repository nobody actually works in. Five or six systems is fine, he argues, as long as a tool can look across them and understand the whole. “Anything else is thirty-year-old thinking in a digital box.”
The room where the number stops mattering
The second conversation took us inside the inspection room. “If something is not planned as an expected document, completeness will never miss it,” said Martin Hausten — a sentence we have been repeating ever since. Martin has spent twenty years at Boehringer Ingelheim, from computer-system validation through GCP auditing to process ownership for TMF management; he spoke with us in a personal capacity, and the views here are his own perspective and do not necessarily reflect the official position of his employer.
Martin has sat on the receiving end of regulatory inspections, and his description of what actually happens in the room quietly reframes the whole debate. Inspectors do not begin with the percentage. They seek to reconstruct what happened, why decisions were made, who was accountable, and whether the evidence is reliable. The TMF becomes decisive when that evidence is incomplete, inconsistent, or slow to retrieve.
The direction of travel makes that sharper, not softer: during remote and on-site inspections, authorities may request direct or facilitated electronic access to relevant systems and records — with a tendency to increase. When inspectors can navigate the evidence themselves, weak metadata, fragmented context and retrieval delays become immediately visible.
When the inspector can log in, “every day is inspection day” — and a percentage does not answer the question they came to ask. Inspection readiness therefore must be a maintained state, not a document-recovery exercise initiated when an inspection is announced.
The monitoring seat
The view from the road comes from Poornima Ellawala, Head of Monitoring at RemediumOne. From a CRA’s perspective, she says, coherence is about whether the documentation reflects what is actually happening at the site. A study-specific checklist matters because every trial has different requirements — a standard checklist may not capture the records that are truly relevant to that study.
During monitoring, her teams are constantly connecting information across source data, site documentation, correspondence and the TMF to understand whether the study is progressing as expected. When those records are incomplete, inconsistent or disconnected, the full picture becomes much harder to see.
TMF management is also time-consuming for both CRAs and site teams, particularly with physical files. Well-managed electronic TMFs, she argues, could reduce that burden and let monitoring visits focus on patients, source data and study conduct, with TMF review handled more efficiently outside the visit.
“Ultimately, it is not just about having all the documents — it is about whether those documents come together to tell a coherent story of the trial.”
The story problem
Which brings us back to Barbara and her telephone book. Barbara Janje-Bode has spent more than fifteen years building, maintaining and improving TMFs across big pharma and biotech — most recently leading document control and archival at a biotech company, and has just completed a professional certification in quality management, GMP and MDR. Her answer to the question is the one no metric can reach.
Years ago she interviewed a candidate who said, almost in passing, “TMF is a kind of art.” She hired her on the spot. The art is arranging a study’s documents, within the TMF Reference model or similar, so that an outsider, new team member, auditor and inspector can pick up the file and understand the story of the trial. The telephone book is the counter-image: every entry present, perfectly ordered, and no story anywhere. You can have every document filed and still fail to give a meaningful picture of the study.
What is missing, she says, is the human perspective and “emotion”. It is an odd word for a GCP conversation, which is precisely why it sticks. She does not mean sentiment. She means context, judgment, nuance and meaning: the understanding of what happened, and why it matters. Studies never go to plan. No two are alike. Sometimes a single sentence changes the relevance of an entire document, a single file note explains the existence of a bunch of seemingly unrelated documents, a small sentence in an email guides to an important decision for the trial. That´s why it´s so important to file and document in time and by someone who lived the trial and can judge what actually matters to the story. Good TMF people, she says, are always storytellers: they ask not only “Where does this document belong?” but “What does it contribute to the story?”
None of this is an argument against standardization. But organizing information is not the same as understanding it: a filing model can tell you where a document belongs, or that it is missing, not whether an unremarkable email carries the context for a major decision. That is why she has come to believe the real challenge was never the technology. The harder task is convincing stakeholders of the purpose, the value and, yes, the art of managing a TMF. But that´s a different story.
What better would look like
When we pushed each of them on what should replace the scoreboard, the answers rhymed.
Eldin would move the quality to the moment the work happens. Close the monitoring visit, confirm its package is complete, lock it — and the quarterly hunt through three months of email disappears. He would even make the gate binding: the visit is not finished until its records are. “If there are gaps in the TMF, you haven’t done your job.” In his words: “you do quality throughout the process.”
Martin draws an analogy with mature quality systems outside life sciences: not because clinical-trial records are interchangeable products, but because those industries distinguish routine variation from failures that materially affect quality. Applied to the TMF, the objective is not to treat every defect as equally important, but to direct controls toward the records and processes that are critical to participant protection, reliable results and trial reconstruction. A minor formatting defect or an isolated short delay may have limited significance, while a missing or unreliable critical record may directly affect the ability to reconstruct the trial. The assessment should consider context, recurrence, criticality and impact — not merely count defects.
And Barbara, asked the question that joins her craft to their engineering: could AI catch what a humans might miss in the pure amount of documents — inconsistencies across documents, across timelines, across the story itself — and sketch a kind of storyboard of the study, while people still do the telling?
Where they part ways
Underneath the agreement sits one real disagreement, and it belongs in the open. Barbara holds that reading a study like a book is a craft belonging to people who lived the trial — and may never fully transfer. Martin emphasises scalable controls, quality gates and risk-based measures: build the gates, measure honestly, accept that perfection is not the goal. Eldin would lock quality into the shape of the process itself, so less of it depends on heroics in the first place. Can the storytelling be engineered? That may be the real question hiding under the metric.
So what does 96% prove?
Four people, four vantage points, one question — and two convergences we did not expect. The first: all three veterans of the file, independently, said look outside life sciences. Eldin and Martin pointed at industries that have managed information and quality at scale far longer than we have; Barbara’s outside world is literature — the file as something that must be readable by someone who was not there. The second belongs to the two men: they pointed, again independently, at the same destination: a file that stops being a warehouse of PDFs and starts being data. ICH M11 points toward a more structured, machine-readable protocol; in time, that structure could support downstream automation — including more dynamic generation of trial-specific record expectations — though that application goes beyond what M11 itself currently requires. And the regulatory language is moving in the same direction: ICH E6(R3) speaks not only of essential documents but of essential records — documents, data and relevant metadata in any format that collectively allow the trial’s methods, decisions, conduct and results to be evaluated. That is a broader test than whether a predefined list of PDFs has been collected. When people this senior describe the same future without comparing notes, that is less an opinion than a bearing.
None of them think completeness is worthless. It tells you the collection happened. What all of them are circling, from the human side, the engineering side, the architectural side and the monitoring seat, is that presence is not the same as coherence. A file can be complete and still not hold together. It can hit 96% and still not read like the story of a trial.
Maybe that is the number worth working on next.
Where do you land on it? We read every reply — and if a question like this is yours, we would like to have you in a future issue.
Contributors
Jürgen Lorenz — Co-Founder and CEO, Menken Trials. Posed the question, and wrote this issue.
Eldin Rammell — Senior Director, GCS Quality Assurance, Cencora (formerly PharmaLex). Thirty-six years in records and information management; long-time secretary/member of the TMF Reference Model steering committee.
Martin Hausten — Global Process Owner, Trial Master File, Boehringer Ingelheim. He advances risk-based TMF governance, inspection readiness, and the effective integration of quality, process, and technology. Contributing in a personal capacity; the views expressed are his own and do not necessarily represent those of Boehringer Ingelheim.
Poornima Ellawala — Head of Monitoring, RemediumOne (Global Monitoring).
Barbara Janje-Bode — TMF and Document Control Expert (GCP compliance, eTMF implementation, inspection readiness). More than fifteen years in clinical research across pharma and biotech, including Hexal; most recently Manager Document Control and Archival at a biotech company. Newly certified in quality management, GMP and medical device QM, and currently open to her next role
