Root cause analysis: getting past the first plausible answer
Root cause analysis works out why something went wrong deeply enough that fixing it prevents a recurrence. Its main difficulty is stopping too early, at an answer that sounds satisfying.
A standard operating procedure is a promise that the work is done the same way twice. Most fail not because they are wrong but because they are unreadable, unfindable or out of date.
A standard operating procedure is a written description of how a recurring task is done, precise enough that two people following it produce the same result. That is the whole definition, and the whole difficulty: precise enough. Too vague and it is a mission statement; too detailed and nobody reads past step four.
The honest reason to write one is repetition with consequences. If the task happens weekly, if getting it wrong costs money or safety or a customer, and if more than one person does it, it deserves a procedure. Everything else deserves a note, or nothing at all.
Write for the least experienced person who will ever legitimately do this task, not for the author. The most common failure in procedures is assumed knowledge: a step that says "reconcile the batch" to someone who has never been told what a batch is. If a step needs a definition, the definition belongs in the procedure, not in somebody's head.
A policy says what must be true — "customer data is never emailed unencrypted". A standard operating procedure says how a task is carried out, end to end. A work instruction goes one level down, covering a single operation in detail for the person performing it. Mixing the three produces a document that is too abstract to follow and too specific to approve. Write the policy once, the procedures per task, the work instructions where the detail genuinely matters.
Every procedure decays, and an out-of-date one is worse than none: it teaches people that the written version is unreliable, and after that they stop looking. Two habits keep the set honest. Give each procedure a review date and treat a missed review as a defect. And make correcting one cheap — anyone who follows a procedure and finds it wrong should be able to comment on it in place rather than mentioning it in passing.
In Ettex, the natural home for this is Ettex Docs: write the procedure, and use threaded comments so the person who found step six wrong can say so against step six, with @mentions to whoever owns it. Version history means you can see what changed and return to any earlier state, so a bad edit is recoverable and an argument about what the procedure used to say is settled by looking. Share links let you point at the procedure from wherever the work happens. Anything that must be signed off — a safety procedure, a regulated one — can go on to Ettex Signature for a real signature with an audit trail, and forms that a procedure tells people to fill in fit Ettex Forms, where submissions land timestamped in one inbox.
What Ettex does not do: there is no SOP module, no library of ready-made procedure templates, no automatic review reminders, and no approval routing between named reviewers. The review date is something you put in the document and diary yourself. If you operate under a standard that demands controlled documents with enforced periodic review and audit-ready approval trails, use a dedicated document-control system.
A written description of how a recurring task is performed, detailed enough that different people following it get the same result. It covers scope, the responsible role, prerequisites, numbered steps, decision points, and what completion looks like.
As short as the task allows — one to two pages for most routine work. Length is the main reason procedures go unread, so split a long one into separate procedures rather than adding sections.
An SOP describes a whole task end to end; a work instruction describes a single operation in the detail the person performing it needs. Many companies need both, and confusing them makes each less usable.
Someone who has watched the task being done, drafting with the person who does it. Procedures written from an org chart rather than from observation describe an imagined process.
Set a review date per procedure — annually is a common default, sooner for anything tied to a tool, a rule or a supplier that changes. Treat a missed review as a defect, not as admin.
Not for most companies — a document tool with comments, version history and a findable location covers it. Dedicated document control pays off under standards that require enforced review cycles and audit-ready approval records.
A standard operating procedure earns its place when a newcomer can follow it without asking anyone. Write from observation, one action per step, name an owner and a review date — and keep it where the work is, not where the documents are.
Root cause analysis works out why something went wrong deeply enough that fixing it prevents a recurrence. Its main difficulty is stopping too early, at an answer that sounds satisfying.
A code of conduct says how people here are expected to behave and what happens when they do not. Its value is not aspiration — it is having decided the hard cases before one arrives.
A project charter names the objective, the boundaries, the sponsor and the person authorised to run the work. Its main use comes months later, when people disagree about what was agreed.