WorkflowClinical Development

Choosing · time and cost

How long does it take to write a clinical trial protocol?

Days to weeks of writing inside a cycle measured in months — and most of the calendar is spent waiting for decisions rather than writing. This guide takes apart where the time actually goes, prices a day of it, and shows the split Clinical Protocol Authoring is designed to change.

Updated · 13 min read

See one case, screen by screen →
Watch with transcript →

Writing a clinical trial protocol takes a medical writer days to weeks of actual writing, inside a development cycle that published programmes measure in months. Those are two different clocks, and conflating them is why protocol timelines are argued about so unproductively. The National Cancer Institute's reformed targets allow 7 months from concept to an activated Phase I–II trial and 10 months for Phase III; a first draft inside that is a matter of weeks of writer time, most of which is spent copying from the last protocol and waiting for someone to answer a question. Clinical Protocol Authoring attacks those two lines specifically, because they are the only ones that can be shortened without reviewing less.

The sample trial, drug and people in the product figures are invented: VLT-302, a Phase 3 study of Veltrimab in plaque psoriasis, written by Lena Ortiz. The timeline and cost figures come from published studies, linked where they appear.

How long does it take to write a clinical trial protocol?

There is no published median for the writing alone, because almost nobody measures it separately from the approval cycle around it. The honest ranges, from the way the work decomposes, are these: a first draft of a Phase 3 protocol in a sponsor template is a few weeks of writer effort spread over longer, because the writer is waiting; a synopsis-to-final cycle on a programme that is not in a hurry runs several months; and the whole distance from concept to a trial open to enrolment is measured against published targets of seven to ten months.

What drives the spread is not the length of the document. A Phase III protocol averages about 116 pages (Getz, Smith and Kravet, 2022), and a competent writer does not take months to produce 116 pages of prose. It is that a good share of those pages cannot be finished until someone else decides something, and the queue for those decisions is the calendar.

What a protocol writer’s days are actually spent on

Decompose a Phase 3 first draft and the largest block is not writing at all. It is adaptation: opening the last protocol on the same drug, copying a section, and changing it line by line while checking that each change has been made everywhere it needed to be.

Estimated working days for one Phase 3 first draft, by what the day is spent on

Copying and adapting approved language
14 days
Writing new text from the synopsis
10 days
Building the schedule of assessments
5 days
Cross-checking sections, synopsis and schedule
5 days
Reading the synopsis and finding the closest past protocol
2 days
Chasing open questions with clinical, statistics and safety
2 days
Our estimate of the split for a Phase 3 protocol from synopsis to first draft, totalling about 38 working days — the by-hand median on the solution's sample dashboard. Orange marks the eighteen days spent on work that produces no new text: finding, copying and re-checking.

Eighteen of those thirty-eight days — copying, finding and cross-checking — produce no sentence that did not exist somewhere already. They are the price of reuse done by hand. Ten days produce genuinely new text from the synopsis, and five build the table everything else refers to.

The two-day figure for chasing open questions is the one that most understates reality, because it counts the writer's effort rather than the elapsed time. Two days of chasing can occupy three weeks of calendar.

Concept to activated trial: the published timelines

The best public evidence on the whole cycle comes from the National Cancer Institute, which set hard timelines for its own trials after an operational efficiency review and then published what happened. "For phase I-II studies and phase III studies, the target timelines are 7 months and 10 months, whereas the absolute deadlines were set at 18 and 24 months, respectively. Trials not activated by the absolute deadline are automatically disapproved" (Abrams et al., Journal of the National Cancer Institute, 2013).

The results are the interesting part. "The initial experience is encouraging and indicates a reduction in development times for phase I-II studies from the historical median of 541 days to a median of 442 days, an 18.3% decrease." Phase III studies, on a smaller sample of twenty-five, showed a 45.7% decrease in median days. On that evidence the institute then cut the absolute deadlines further, to 15 and 18 months.

A cancer trial from proposal to activation, against the targets set for it

Historical median, phase I–II541 days
Development, proposal to activation541 days
  • Development, proposal to activation541 days
After the timeline reform, phase I–II442 days
Development, proposal to activation442 days
  • Development, proposal to activation442 days
Target, phase I–II7 months
Target timeline7 months
  • Target timeline7 months
Target, phase III10 months
Target timeline10 months
  • Target timeline10 months
Absolute deadline, phase III24 months
After which a trial is automatically disapproved24 months
  • After which a trial is automatically disapproved24 months
National Cancer Institute trials after the Operational Efficiency Working Group reforms. Lanes are drawn to scale in days; targets converted at 30.4 days a month. On the strength of these results the absolute deadlines were later reduced to 15 and 18 months.Source: Abrams et al., implementation of timeline reforms speeds initiation of NCI-sponsored trials (2013)

Two lessons travel out of oncology. First, a historical median of 541 days — about eighteen months, against a target of seven — is not writing time; the gap is review, negotiation and queue. Second, the thing that moved the median was not a new drafting tool. It was a deadline with a consequence attached, plus new staff positions to monitor progress and protocol-tracking websites to watch the clock.

What a day of delay costs

The figures most often quoted for a day of delay are decades old and, in the words of the researchers who replaced them, "outdated and based on anecdotal evidence". Tufts CSDD rebuilt them from data: 645 drugs launched since 2000 and 409 clinical trial budgets. "The estimated direct daily cost to conduct a clinical trial is approximately $40,000 per day for phase II and III clinical trials, with those in respiratory, rheumatology, and dermatology having the highest relative daily direct costs" (Smith, DiMasi and Getz, Therapeutic Innovation & Regulatory Science, 2024).

What a day of delay costs, rebuilt from 409 trial budgets and 645 launches

$40K

approximate direct cost of a day of a Phase II or Phase III trial

$500K

approximate value of a day of lost prescription drug or biologic sales

$80–100K

how much average sales per day have fallen each year

Smith, DiMasi and Getz, peer-reviewed analysis of 645 drugs launched since 2000 and 409 clinical trial budgets. Respiratory, rheumatology and dermatology trials had the highest relative daily direct costs.Source: Smith, DiMasi and Getz, new estimates on the cost of a delay day in drug development (2024)

Two cautions before anyone multiplies. The daily figure is the direct cost of conducting a trial — it is not the cost of a day spent writing the protocol before approval, when the trial is not yet burning site and vendor money. And the same study is corrective about the other number people reach for: a day of lost prescription sales is worth roughly $500,000, far below the multi-million figure long repeated, and average sales per day have been falling by roughly $80,000 to $100,000 a year.

What the daily figure does price honestly is anything that extends the trial after it starts — most obviously an amendment that adds a visit, a country or a cohort. Note too which therapeutic areas sit at the top of the daily cost ranking: respiratory, rheumatology and dermatology. A dermatology Phase 3 like the sample trial is at the expensive end.

The rework nobody puts in the plan

Every protocol timeline is drawn as though the protocol will be written once. Published data says otherwise. Tufts CSDD found that a majority of protocols — 57% — took at least one substantial amendment, that nearly half of those amendments were judged avoidable, and that the median direct cost of implementing one was $141,000 for a Phase II protocol and $535,000 for a Phase III protocol (Getz et al., 2016).

An amendment is not only a cost; it is a second timeline. The change has to be written, classified, submitted or notified, assessed, approved by ethics committees and adopted site by site — and in the European Union a substantial modification may not be used until it has been authorised in that Member State. A protocol finished four weeks early and amended twice has not saved anybody four weeks. Which changes need which route, and why the classification is expensive, works through that sequence.

Why copying from the last protocol is the slow option

Copying feels fast because the first hour is fast. The cost arrives later, in two forms. The first is the change that was made in six places and needed seven: a study whose primary endpoint moved from Week 12 to Week 16 can carry a schedule in which Week 16 has no efficacy ratings at all, which is exactly the finding sitting at the top of the sample trial's draft. The second is the cross-check, five full days of reading a hundred pages against a nine-page synopsis and a table, looking for the one number that did not move.

Copying the last protocol, and assembling from sourced approved language

Copy and adapt by handAssemble with the source kept
Finding the right past protocol

Copy and adapt by hand: Two days of asking colleagues and searching a document store, then reading to judge the fit.

Assemble with the source kept: Candidates ranked by population, phase and therapeutic area, with the match stated — in the sample trial 92%, 71% and 54%.

Knowing what changed

Copy and adapt by hand: A manual diff, or trust. Changed words are invisible once the document is saved.

Assemble with the source kept: Every adapted passage marks each changed word; every reused passage keeps its document, version and page.

Catching the thing you missed

Copy and adapt by hand: Five days of cross-checking, late, by the person least able to see their own omissions.

Assemble with the source kept: Checks run on the draft before any reviewer opens it — endpoint timepoints, objectives against estimands, windows, dose, abbreviations.

Reusing from a protocol that was later amended

Copy and adapt by hand: Nothing tells you. The amendment lives in a different document.

Assemble with the source kept: The source is flagged as later amended before its language is reused.

What the reviewer receives

Copy and adapt by hand: A document, and an assurance that it was checked.

Assemble with the source kept: A draft where each passage is marked reused, adapted or new, and every source opens at its page.

Left column describes the by-hand method the estimate above is built on. Right column describes Clinical Protocol Authoring on the invented sample trial, where 58% of the first draft was reused from final protocols with its source.

A nine-page synopsis, read into the fields a draft needs

The synopsis view in Clinical Protocol Authoring for the sample trial VLT-302. A nine-page protocol synopsis is shown with passages highlighted where the agents drafted from them: the endpoints paragraph giving co-primary PASI 90 and sPGA 0/1 response at Week 16, the population paragraph defining adults with chronic plaque psoriasis for at least six months, and the intervention paragraph giving 150 mg subcutaneously at Weeks 0 and 4 then every 12 weeks with contraception until 20 weeks after the last dose. A right-hand panel lists what the agents draft from — the synopsis document read into 31 fields, the sponsor template version 3.2 with its fourteen sections, and a library of 38 final protocols and 412 approved passages — then ranks the closest approved protocols at 92%, 71% and 54% match.
Invented sample data. The two days a writer spends finding and judging the closest past protocol is the first line in the estimate above, and the one that collapses most completely.

Waiting for decisions: the queue inside the calendar

A synopsis is nine pages and a protocol is a hundred; the gap is filled with decisions the synopsis did not make. Four of them in the sample trial are typical of the genus, and each one blocks a different section.

The decisions a synopsis leaves open, who answers each, and what waits

The open questionWho answers itWhat is stalled while it is open
Should prior biologic use be capped? The earlier study excluded participants with more than two prior biologics; the synopsis says nothing.Dr. Maya Chen, clinical leadSection 5, the eligibility criteria — and with them the screening row of the schedule and the sample size assumptions.
Which testing order for the four key secondary endpoints? The synopsis lists them but gives no multiplicity strategy.Sam Patel, biostatisticianSection 10, statistical considerations, and the estimand table in Section 3.
Is a data monitoring committee needed for a placebo-controlled trial with 52 weeks of exposure?Dr. Priya Raman, medical monitorSection 11, trial oversight, and the safety reporting arrangements that follow from it.
Is the Week 16 switch from placebo to active treatment blinded?Dr. Maya Chen, clinical leadSections 4 and 6, trial design and intervention, and the Week 16 column of the schedule.
Invented sample data from the VLT-302 draft. Each is one question for one named person, with the answer written back into its section with the name and the time. Counted as writer effort these are two days; counted as calendar they are usually the longest single thing in the schedule.

This is why "the protocol is nearly finished" can be true for a month. The remaining work is not writing; it is four people, three of whom are on other trials, and none of whom has been asked a question small enough to answer between meetings. A question that names the decision, states what the previous study did and what the draft currently assumes is answerable in one minute. A request to "review the draft" is not.

What a realistic timeline looks like now

With the copying, the searching and the cross-checking taken out, and the open decisions turned into four named questions, the writing side of the estimate collapses from about thirty-eight working days to about four. That is a design target rather than a measured customer result, and it covers synopsis to a checked first draft — not to an approved protocol, which still contains every review the sponsor's procedures require.

The same Phase 3 first draft, estimated by hand and against the design target

What the day is spent onBy hand (days)Design target (days)
Reading the synopsis and finding the closest past protocol20.1
Copying and adapting approved language, section by section141
Writing new text from the synopsis101.5
Building the schedule of assessments50.5
Cross-checking sections, synopsis and schedule50.5
Chasing open questions with clinical, statistics and safety20.4
Total working days384
Our estimate for the by-hand column; the right-hand column is the solution's design target, not a measured customer result. On the sample dashboard the by-hand median is 38 days and the median days to first draft across eighteen protocols is 5.

Note what has not moved. Writing genuinely new text from the synopsis still takes real time, because it is real work. The four days are bought almost entirely from the eighteen that produced nothing new, plus the two spent chasing people who can now answer in writing.

Days to first draft across a portfolio, not one protocol

A dashboard in Clinical Protocol Authoring with four summary tiles: visits over the burden limit 23, with 20 brought under before sign-off; median days to first draft 5 days, was 38 days by hand; approved language reused 58% with its source; inconsistencies caught 64 before team review. A bar chart titled median days from synopsis to first draft shows five quarters falling from 38 to 22, then 7, 4 and 3.5, with the first bar labelled written by hand and the rest drafted with agents and approved by writers. A second panel shows where the draft text came from — reused verbatim with source 41%, adapted with changes marked 17%, new from the synopsis 29%, template standard text 13% — across eighteen protocols.
Invented sample data. The useful number on this screen is not the median but the shape of the second panel: 41% of the drafted text was reused verbatim from a final protocol with its source — the share of the draft where the copying days in the estimate above used to go.

Can it be faster without being riskier?

Only if the time comes out of the right places. Every hour saved on finding, copying and re-checking is an hour that was never adding judgement, and removing it makes the draft more reliable rather than less, because those are precisely the steps where human attention fails. Every hour "saved" by shortening review is borrowed against an amendment.

There is a second, slower effect worth stating. Protocols have been getting heavier: mean total procedures per Phase II–III protocol rose 67.3% between 2009–2012 and 2017–2020 (Getz, Smith and Kravet, 2022), and the average duration of each visit rose with them, "in large part due to the volume of procedures performed per visit" (Dirks et al., 2025). A protocol that is quick to write and heavy to run is not a win. Measuring that weight while the schedule is still being drawn is the other half of a faster cycle.

Eight ways to take real days out of protocol development

  1. Put a deadline on activation, with a consequence

    An intervention with published evidence behind it: NCI cut phase I–II median development time from 541 to 442 days by setting targets, absolute deadlines and automatic disapproval.

  2. Make the open decisions explicit on day one

    List every question the synopsis does not answer, name the person who owns each, and send them as questions rather than as a draft to review.

  3. Start at the schedule of assessments, not Section 1

    The table constrains eligibility, intervention, assessments, safety and statistics. Writing prose before it is settled guarantees rewriting.

  4. Reuse with the source attached, not by copying

    A passage that carries its document, version and page can be checked in seconds; a pasted one has to be re-verified from scratch.

  5. Run the consistency checks before the first human review

    Reviewers should spend their attention on design, not on finding the Week 12 that should have become Week 16.

  6. Review in parallel, by section, with named owners

    Sequential review of a whole document is the slowest possible arrangement and the one most likely to produce contradictory comments.

  7. Score patient burden before the draft leaves the building

    A visit that sites reject after enrolment costs an amendment; the same finding before approval costs an afternoon.

  8. Measure days to first draft, and the amendments afterwards

    Either number alone is misleading. A team that is getting faster and amending more has moved work, not removed it.

The first item is drawn from the published NCI timeline reform; the rest are our recommendations, in the order they usually pay off.

Ten protocols, and where each one actually is

A protocols list in Clinical Protocol Authoring showing ten open drafts and two final ones. Each row gives the study and compound, the phase, a five-step stage bar reading synopsis, drafting, writer review or team review, sections approved out of fourteen, checks still to fix, the owner and a due date. Studies range from Phase 1b to Phase 3 across three compounds, with owners Lena Ortiz and Jonas Weber; one Phase 2 study shows 14 of 14 sections approved and clear checks in team review, while others show 0, 5, 6, 8, 9 and 11 sections approved with one to three checks to fix.
Invented sample data. A protocol's real position is the sections-approved column and the checks column together: a draft at 11 of 14 with one check open is close, and one at 14 of 14 with three open is not.

Protocol timelines shorten when the mechanical work stops being done by people and the decisions stop waiting in an inbox. To see the synopsis, the schedule and the checks on one screen, start from Clinical Protocol Authoring.

Questions people ask

How long does protocol development take from concept to final?

Months rather than weeks, and the published benchmarks are from oncology: NCI set targets of 7 months for phase I–II and 10 months for phase III from proposal to an activated trial, against a historical phase I–II median of 541 days. After the reforms that median fell to 442 days. Industry cycles vary widely, but the shape is the same — writing is a small share of the elapsed time.

Why do protocols take so long to write?

Because most of the work is not writing. In our estimate of a Phase 3 first draft, eighteen of about thirty-eight working days go on finding the closest past protocol, copying and adapting its language, and cross-checking the result — none of which produces a sentence that did not already exist. The rest of the calendar is spent waiting for decisions the synopsis did not make.

How much does a day of delay cost in a clinical trial?

About $40,000 a day in direct costs for a Phase II or Phase III trial, according to a 2024 peer-reviewed analysis of 409 clinical trial budgets, with respiratory, rheumatology and dermatology trials the most expensive per day. The same study puts a day of lost prescription drug or biologic sales at roughly $500,000, and finds that figure falling by about $80,000 to $100,000 a year.

How many people review a protocol?

In the sample trial, five: the medical writer approves every section, then each section goes to its named reviewer — the biostatistician for objectives and estimands, the clinical lead for population and intervention, the clinical operations lead for the schedule and procedures, the medical monitor for safety — before a full team review and the head of medical writing's signature. The number matters less than whether they review in parallel by section or sequentially over the whole document.

Can a protocol be written faster without more risk?

Yes, if the time comes out of copying, searching and cross-checking rather than out of review. Those are the steps where human attention is weakest, so automating them makes a draft more reliable. Time taken out of review has the opposite effect, and tends to come back as an amendment at a median direct cost of $535,000 for a Phase 3 protocol.

What slows a protocol down most often?

An unanswered question that blocks a section. In the sample trial four decisions the synopsis left open — a cap on prior biologic use, the testing order for key secondary endpoints, whether a data monitoring committee is needed, and whether the Week 16 switch is blinded — each hold up a different section and the parts of the schedule that depend on it. Asking them as four named questions rather than as one review request is the single cheapest improvement available.

Clinical Protocol Authoring.
See it on your own files.

An agentic solution from Botminds. Trial protocols drafted in your template around the schedule of assessments, with every visit scored for patient burden. Book a demo, or explore how it works.