Approval workflow

What an approval workflow is

An approval workflow is the agreed route a draft takes from the person who wrote it to the published page. It names the stages, names a person at each stage, says what that person is allowed to comment on, gives them a deadline, and states what happens if the deadline passes or two reviewers disagree.

It is a small document. One page and a diagram is usually the whole of it. Its value is that it removes the two things that actually kill content in a small business: a draft sitting unread on one desk, and a fourth reviewer arriving at the end with an objection about something settled in week one.

The page it is most often confused with is content governance. Governance is about pages that are already live: who owns them, who reviews them, on what cycle. An approval workflow is about work in progress. A piece leaves the workflow the moment it publishes and enters the governance register that same day.

It is also not approval management, which is me doing the chasing for you each month. The workflow is the design. Management is the running of it.

When you need one

Three situations. Drafts are finished but not published, and nobody can say who is holding them. Everything routes through the owner, who is the busiest person in the business, and the queue moves at the speed of their evenings. Or approval has crept: what used to be one reader is now four, each adding a pass, and a two-day edit takes five weeks.

If two people write and publish between them and nothing waits, you do not need a workflow. You need a written bar for what good enough looks like, so the check is a checklist rather than a meeting.

What goes in it

• The stages. Usually four: write, edit, review, publish. Any business with more than five stages is describing a problem rather than a process.
• The named roles. A person, not a department, at each stage, with a named substitute. “Marketing reviews it” is how a draft gets lost.
• The scope of each review. What each reviewer may raise, and what they may not. A technical reviewer checks accuracy and is not asked about the headline. This is the single line that saves the most time.
• Deadlines and the default. Two working days at each stage, with a stated default when the time passes: for most pieces, silence approves. For the short list of high-risk pieces, silence does not, and the deadline escalates instead.
• The deadlock rule. When two reviewers disagree, the piece goes to one named person who decides within one day and whose decision ends it. Without this, a draft can be argued over indefinitely by people who are all being reasonable.
• The fast lane. The categories of piece that skip review entirely, so routine work is not dragged through machinery built for the risky exceptions.

What you get

The route map. One page: the stages, the names, the deadlines, the fast lane and the deadlock rule, drawn so that it can be understood without reading the text underneath.

The role cards. A short paragraph per role saying what that person is responsible for and, more usefully, what they are not. Reviewers are given these individually.

The wording. The exact sentences for the default-approve rule, the escalation and the deadlock decision, ready to paste into whatever tool you use to pass work around.

A pilot and a review. The workflow run on three real pieces, then half an hour to fix whatever the first three pieces exposed, because the first draft of a process is always slightly wrong.

A worked example

An illustration, not a client. A Halifax company delivering software training to public-sector clients publishes about four pieces a month. Drafts go to the head of training for accuracy, the managing director for tone, and a part-time marketing contractor for the site. Nothing is written down, so every piece goes to all three, each waits to see what the others said, and the average draft takes five weeks to publish. Two pieces have been in the queue since spring.

The new route has three stages. The head of training reviews accuracy only, in two working days, and silence approves. The contractor publishes. The managing director sees only pieces in two named categories: anything about a public-sector client, and anything that quotes a price. Everything else skips them. Disagreements go to the managing director and are settled in a day. The two stalled pieces publish that fortnight, and the monthly count goes up without anyone writing faster.

How it runs

1. A half-hour call. What you publish, who currently has to see it, and where the last few pieces got stuck. No charge.
2. A fixed price in writing. Agreed before anything starts, based on how many roles and content types are involved.
3. Tracing. The last ten pieces followed from first draft to publication, with the actual delay at each step recorded. Two to three working days, and it is usually the part that changes people's minds.
4. The draft route. Agreed with you first, then taken to each named reviewer individually, because a reviewer who has not agreed to a deadline will not meet it.
5. The pilot. Three real pieces through the new route, then the final version. Two to three weeks from the first call.

What it costs

Quoted as a fixed price in CAD after the half-hour call and agreed in writing before the work starts, because it depends on how many reviewers there are and how tangled the current route is. It does not change afterwards without your agreement. The pricing page sets out the four ways of working.

What happens next

A route works better when reviewers are checking against something written, so most people pair it with a definition of good enough to publish. Once pieces are shipping, the pages need owners, which is the ownership register. And if the honest answer is that nobody internally will chase it, having the chasing done for you is the alternative to writing a process that nobody runs. None of that is assumed. The workflow is complete once it is agreed and piloted.