Two people in the same lab, running the same protocol, on the same instrument, get results that differ by more than they should. This is so common it is usually treated as noise. It is not noise. It is a documentation problem wearing a lab coat.
The protocol is not the thing being shared
What is written down is the protocol. What is actually shared, in a functioning lab, is the protocol plus a large body of unwritten practice: how long "briefly vortex" is, which of the two water baths is trusted, whether you pre-warm the medium, what everybody does when the plate reader gives a negative value.
New people do not inherit this by reading. They inherit it by standing next to someone, which means it propagates unevenly and mutates as it goes.
Write down the disagreements, not the agreements
The instinct when a lab decides to standardise is to write more protocol. This rarely helps, because the parts everyone already does the same way are the parts that get written, and the parts that vary are the ones nobody thinks to mention.
A faster route: ask two people who both run a protocol to each describe one step in detail, separately, and compare. The differences that surface in ten minutes are usually the ones that have been costing you reproducibility for a year.
Share the artefacts, not just the instructions
A protocol says "prepare the master mix". The artefact is the actual recipe, with the reagents, the concentrations and the overage, that everyone opens rather than re-derives. When each person re-derives it, each person makes their own small decisions, and those decisions are exactly the variability you are trying to remove.
The same is true of plate layouts, dilution series and buffer recipes. A shared layout that everyone opens is a stronger standard than a paragraph describing the layout, because it removes the step where interpretation happens.
Make the shared version obviously the shared version
Any system where the canonical copy is indistinguishable from a personal copy will drift. Folders help here more than rules do: one place that is understood to hold the group's versions, and personal work kept somewhere that nobody mistakes for it.
Keep the bureaucracy proportional
Standardisation has a real cost, and a lab that standardises everything stops being able to try anything. The test is whether a difference between two people could change a result. If it could, write it down. If it could not, leave people alone.