Rebuilding an operations manual: six volumes, one voice

01

What six volumes actually looked like

Consider a manual that has grown into six volumes through successive additions. The first task is to identify which instructions still govern the work, which conflict and which have no owner.

Clinical standards was one. Training and competency was another. Front of house and patient handling was a third. Then stock and procurement, then daily reconciliation, then a final volume that was really a forms library with instructions attached to the back of each form.

Each volume had been written by a different function, in a different year, by someone competent, for a reason that made sense at the time. None of that was a mistake. The mistake was that no one had ever been given the job of reconciling them, and nobody owns a problem that is nobody's job.

The failure mode was quieter than people expect. Staff did not ignore the manual. They used it, and got different answers.

Two volumes described the same consent step in a different order. One put the photographs before the signature, one put them after. Both documents were signed off. Both were current. A site manager following either one in good faith was compliant with a document and out of step with a colleague two hours away. When that surfaced in a review, the conversation was never about the patient. It was about which volume outranked the other, and nobody could answer.

02

Length was not the problem. Voice was.

The instinct when you inherit six volumes is to cut. That instinct is mostly wrong, and it is worth understanding why before you spend a quarter on it.

A manual should help someone confirm the approved action at the point of work. I would test whether the answer can be found and understood in a realistic scenario. The test matters more than the number of sections the document contains.

Which means the damage from six voices is larger than the damage from six hundred pages. A long document written by one mind teaches you how to read it. After three sections you know where the trigger sits, where the escalation sits, what a bold line means. Six documents written by six functions teach you nothing transferable. Every section is a new dialect.

A manual in six voices is not one standard written six times. It is six standards, and people will use whichever one is nearest to hand.

03

The five pass rebuild

This is the sequence I used, and it is the part a reader can start on a Monday. The rebuild runs in five passes, in this order: inventory, contradictions, one pen, one owner per section and retirement. One senior person owns the whole sequence. Each section owner holds a review date.

Pass one, inventory. One sheet listing every document anyone in the network treats as binding. Author, date last changed, and whether a single person has opened it this year. Not what should be binding. What is.

Pass two, contradictions. Read the inventory looking only for places where two documents instruct differently on the same task. Fix nothing yet. Just list them. This list is the thing that gets you the mandate, because it is the only artefact that makes the problem undeniable to people who like the current manual.

Pass three, one pen. One person writes the final text of everything. Functions supply the content, argue the substance and sign off on accuracy, but they do not supply sentences. This is the pass everybody resists and it is the whole rebuild. One voice is a structural decision about authorship, not a style preference.

Pass four, one owner per section. A name, not a department. Departments do not read calendars, people do. Each owner gets a review date and the standing right to propose changes, which is what stops the document ossifying the day it ships.

Pass five, retirement. Publish the list of what is now void, and remove it from the shared drives, the binders and the folders on site desktops. A rewrite that leaves the old volumes reachable has not replaced anything. It has added a seventh.

04

What actually changed

I am going to be careful here, because the honest outcomes are qualitative and I would rather say less than dress it up.

I would measure whether teams can find the same approved answer and whether disagreements now concern a specific rule rather than conflicting copies. A common format is useful only if it improves that work.

New sites stopped writing local annexes. When a site cannot get a straight answer from the centre it writes its own note, and those notes are where network consistency quietly dies. That behaviour reduced noticeably.

Test onboarding with the revised manual and record where a new manager needs help. Compare those findings with the original usability problems. Do not claim a faster induction simply because the document looks more consistent.

What did not change: the total page count barely moved. I had expected a dramatic reduction and did not get one. Consolidation removes duplication, not obligation.

05

The document was never the intervention

A searchable platform does not remove the need for training, ownership and current content. I would evaluate those separately before deciding whether a software change is necessary. The first question is whether the team can use the guidance it already has.

Read that as a warning about where you put your effort. A rewrite is necessary and it is not sufficient. If the rebuild ships without teaching, without a cadence for review, and without somebody senior who owns the whole document, the six volumes reassemble themselves within two years under different names.

The Healthcare Financial Management Association makes the same point from the administrative side, recommending that policy management sit with one senior individual or office rather than being distributed, and listing inconsistent style, outdated content and difficulty finding things among the recurring failures. That is a description of how manuals decay when authorship is shared and ownership is not.

06

What I would do differently

Start with the contradiction list. Rewriting an uncontroversial section may improve its prose without resolving the decision that blocks the team. I would settle those conflicts with the relevant owners before polishing the rest.

Cap the first release. Ten procedures, the ten people actually use, published properly and taught properly. I tried to land the whole thing and the size of it became the story.

Write the retirement list before the new text rather than after. If you cannot say what dies, you have not finished deciding what lives.

And name the owners publicly. I left some sections owned by a function because it avoided a conversation, and every one of those sections was the first to go stale.