There is a gap between the methods section of a paper and the thing a person can follow at the bench. The methods section is a compressed account written for reviewers. The protocol is an instruction written for someone tired, alone, and holding a pipette.
Most labs have the first and assume it is the second. Then a new student runs it, gets nothing, and spends three weeks discovering that the protocol assumed a step everyone in the lab does without thinking.
Anyone writing down a procedure that another person will run — for a new student, a collaborator, or your own future self. Works on paper, in a shared document, or in whatever system your lab uses.
The test: could they run it without asking you?
That is the whole standard. Not “is it accurate” — accuracy is easy and most protocols pass. The question is whether someone can complete it without needing a conversation.
The way to find out is not to reread it. It is to watch someone run it once and write down every question they ask. Those questions are the protocol's actual defects, and there are usually between five and fifteen of them.
The five things that always go missing
1. The deviation that became the method
The written protocol says thirty minutes. Everyone in the lab does forty-five, because thirty never worked. It is in nobody's notes, and it is the single commonest reason a protocol fails in someone else's hands.
2. What it should look like at each stage
“The pellet should be small and translucent, roughly the size of a grain of rice.” Without this, a person cannot tell whether to continue or start again — and will usually continue, wasting the rest of the day.
3. Where the timing is rigid and where it is not
Some incubations tolerate an hour's drift; others fail at five minutes. A protocol that presents every step with equal urgency teaches nothing, so the reader either rushes everything or nothing.
4. What was actually used, not what the catalogue says
Supplier, catalogue number and lot for anything that has ever behaved differently between batches — antibodies above all. “Anti-X antibody” is not reproducible, and this is the detail a reviewer is most likely to ask for.
5. The setup nobody counts as a step
Which centrifuge, at what temperature, pre-cooled from when. The buffer that has to be made fresh. The thing that must be booked two days ahead. These are not steps to someone who does them automatically, and they are the whole morning to someone who does not.
A format that fits on two pages
Long protocols do not get read at the bench. This is the shape that survives.
1 · Purpose — one sentence, and when you would use this rather than something else.
2 · Time and shape — total time, hands-on time, and any point where it can be paused overnight.
3 · Before you start — equipment to book, buffers to make fresh, anything to pre-cool.
4 · Materials — supplier, catalogue and lot for anything batch-sensitive.
5 · Steps — numbered, one action each, with the checkpoints written in.
6 · What good looks like — and the two commonest ways it goes wrong.
7 · Version and contact — date, who wrote it, who to ask.
Blocks two, three and six are the ones usually absent, and they are the ones that decide whether the protocol works for a stranger.
Writing the steps
The last rule — say where the output goes — is where protocols and records meet. woodle.cloud keeps a result attached to the run and the protocol version behind it, so “which version did we follow” stops being a memory question. The format above is worth using on paper regardless.
See how it works →Versions, without ceremony
Protocols change, and the change is usually the interesting part. A protocol with no version is a protocol you cannot cite in a methods section eighteen months later.
The minimum that works: a date and a one-line note of what changed and why. “2024-06-11: incubation 30 → 45 min, thirty was giving partial digestion in the new buffer lot.” Keep old versions rather than overwriting — when a result from last year looks odd, the first useful question is which version produced it.
Where to keep them
Wherever people will actually look, which is usually not where the PI would prefer. A printed folder by the bench beats a perfect document in a system nobody opens. Two rules regardless of medium:
- One canonical copy. The moment there are two, they diverge, and nobody knows which is current.
- Anyone can propose a change; one role approves it. Protocols that only the PI may edit stop reflecting what the lab does within a year — and the real method moves into people's heads, which is where this guide started.
The seven blocks and the step rules are free to copy into your own lab documentation. No attribution needed and no signup. If you change one thing, write down the deviation that became the method.