Nick QuirkMost agencies do not have an editorial process. They have a spreadsheet, a shared drive, and a managing editor who remembers everything. That works at four articles a month. At twenty-five, it produces missed deadlines, three versions of the same draft, and a client asking why the blog post they approved in March still is not live.
A real editorial process is a sequence of stages with named owners, fixed durations, and a definition of done at every handoff. Below is the version that holds up across multiple clients, freelance writers, and the inevitable week when your best editor is on leave.
Why editorial work breaks down at agencies
Content is unusual among agency deliverables because it passes through more hands than anything else you sell. A technical audit is produced by one specialist and delivered. A 1,600-word article touches a strategist, a writer, an editor, an SEO reviewer, a designer for images, an account manager, and the client — seven handoffs, each one a chance for the piece to stall.
Three failure patterns account for nearly all of it:
- Vague briefs. The writer guesses at intent, the editor rewrites half the draft, and you pay twice for the same 1,500 words.
- Invisible queues. A draft sits in someone's inbox for six days and nobody notices until the client asks.
- Unbounded client review. No SLA, no round limit, so a piece scoped at eight hours consumes fourteen.
Every stage below exists to kill one of those three.
The nine stages, with realistic durations
These are working numbers for a 1,500-word SEO article at a mid-sized agency. Adjust for thought leadership, which needs more SME time, or for high-volume programmatic content, which needs less.
1. Topic approval (0.5 days)
Topics come from a quarterly content plan tied to keyword research, not from a monthly scramble. Approve a quarter's worth at once — twelve topics, target keyword, search intent, funnel stage, and a one-line angle for each. This single change removes a recurring two-day delay from every piece, because you are no longer negotiating what to write while the calendar burns.
2. The brief (1–2 days)
The brief is the highest-leverage document in the entire editorial process. A good one takes a strategist 45–60 minutes and saves the editor two hours downstream. It must contain:
- Primary keyword plus 4–8 secondary terms, with actual search volumes
- Search intent and the format that satisfies it (comparison, how-to, definition)
- Target word count as a range, not a hard number
- An H2/H3 outline with a sentence describing what each section must answer
- Three competing URLs and one line on what each gets wrong
- Internal links to include, by URL
- Required proof: data, screenshots, customer quotes, SME interview
- Voice notes specific to this client, not generic style-guide language
If you use AI to accelerate outline generation, keep the strategist in the loop for intent and angle — the parts a model gets plausibly but generically wrong. Our guide on structuring AI prompts for marketing deliverables covers how to get briefs that are actually usable rather than restatements of the SERP.
3. Assignment and scheduling (0.5 days)
Assign by capacity, not by who is best. Track each writer's committed word count for the week. A strong freelancer producing 6,000 words weekly cannot absorb a fourth 1,500-word piece just because they are good. Publish the due date to the writer at assignment, and make the due date the draft delivery date, never the publication date — the two are 7–10 days apart.
4. Draft (3–4 days)
Give writers a full working day more than they think they need and hold the line on the deadline. Two rules that pay for themselves: no draft is accepted without the brief's outline honoured (deviations must be flagged in a comment), and no draft is accepted without sources linked inline.
5. Substantive edit (1 day)
This is structure, argument, and accuracy — not commas. The editor asks: does this answer the query in the first 100 words, is every claim supported, is anything here that a competitor could not also have written? Budget 60–90 minutes for a 1,500-word piece. If it consistently takes three hours, the brief is failing, not the writer.
6. SEO and QA gate (0.5 days)
A checklist stage, and the one most agencies skip when they are behind. Title tag under 60 characters, meta description written, H1 and H2 hierarchy clean, primary keyword in title/H1/first paragraph, internal links live and correct, images compressed with alt text, schema where applicable, no broken external links, client's product named correctly. Ten minutes with a checklist prevents the email that starts "Hi, small thing, but…". If you are running volume, semi-automating this gate is worth the setup time — see AI-assisted QA for what to automate and what to keep human.
7. Client review (3–5 days, SLA'd)
The stage that destroys your schedule if it is not contractually bounded. Put this in the scope of work: five business days for feedback, one consolidated round, auto-approve if no response. Send one link, not a thread. Tell the client explicitly what you want reviewed — factual accuracy and product claims — and what is already settled — structure, keyword targeting, headline. Without that framing you get a marketing manager rewriting your H2s.
Track review turnaround per client. When a client averages nine days against a five-day SLA, that number belongs in the QBR alongside the publishing gap it caused.
8. Publication (0.5 days)
Someone owns the CMS upload, and it is a named person, not "the client's dev team, eventually". Stage it, check rendering on mobile, submit the URL for indexing, and update the internal linking on two or three older posts to point at the new piece. That last step takes eight minutes and is skipped 90% of the time.
9. Post-publish tracking (ongoing)
Set a 90-day check on every published URL: impressions, average position, clicks, conversions if you can attribute them. Feed the results back into the next quarter's topic approval. Without this loop, your editorial process produces output but not learning.
Making the process visible
Stages only work if everyone can see where each piece sits. Build a board with one column per stage and enforce two rules: one owner per card at all times, and a WIP limit on editing. Editing is almost always the bottleneck; capping it at, say, six pieces forces the queue to surface upstream instead of hiding in someone's inbox.
Add cycle time tracking. If your target is 12 business days brief-to-publish and your actual median is 21, the board will show you exactly which stage owns the extra nine days. In practice it is client review 70% of the time and drafting the rest.
Whatever tool you use — and there are many decent options; we maintain a comparison of PM tools for agencies if you are choosing — the requirement is the same: per-stage ownership, due dates that cascade, and time logged against the piece so you know whether that £600 article actually cost £600. Agencies running PeakKR typically template the nine stages as a phase set and clone it per article, which removes the setup tax on every new piece.
Costing the process honestly
Run the numbers once. A 1,500-word article at typical rates: brief 1 hour at £70, writing 5 hours at £45, substantive edit 1.5 hours at £60, QA 0.5 hours at £40, PM and client handling 1 hour at £70, upload 0.5 hours at £35. That is roughly £480 of delivered cost before overhead. If you sell it at £600 and your review rounds routinely run to three, you are losing money on content you believe is profitable.
