In most content teams, a draft’s fate is decided by a mood. The editor reads it, feels mostly good about it, fixes a few sentences and publishes. Or feels uneasy about something they can’t name, sends it back with “needs work” and moves on. Either way, the decision lives in one person’s head and leaves no record.
That works until it doesn’t: until a statistic turns out to be invented, a client discovers their product was described wrongly, or a health claim slips into a post about kitchen gadgets. Then everyone wants to know who approved it and why, and the honest answer is “it seemed fine.”
A publish gate replaces mood with rules. And the most useful gate has three verdicts, not two: Publish, Hold and Escalate. This article explains each one, the rules that trigger them and why the third verdict is the one most teams are missing.
Why isn’t “publish or revise” enough?
Because it forces two very different situations into the same bucket. “Needs work” lumps together a draft with a missing link and a draft making a legal claim the writer isn’t qualified to assess. The first needs ten minutes from the writer. The second needs a different person entirely.
When both get labeled “revise,” the writer often tries to fix the second kind alone, usually by softening language until the problem is hidden rather than solved. A third verdict makes that impossible. It says: this isn’t yours to clear.
What do three-way decisions look like elsewhere?
Systems with three outcomes show up wherever a binary choice would hide something important.
The three-color traffic light. Early traffic signals had only stop and go. William Potts, a Detroit police officer, is widely credited with developing a three-color signal around 1920, adding a warning phase between them. The middle state made the transition safer for everyone.
Scotland’s three verdicts. For centuries, Scottish criminal courts could return “guilty,” “not guilty” or “not proven.” Whatever one thinks of that third verdict, it shows a system explicitly recognizing that “insufficient evidence” is a different state from “innocent.”
The andon cord. In the Toyota Production System, any worker on the line can pull a cord or press a button to signal a problem and summon help. The line doesn’t just pass or fail parts. It has a built-in way to stop and call someone with more authority.
Publish, Hold and Escalate map neatly onto these: go, not yet and get someone else.
What does each verdict mean?
Publish
Every gate is clear. Claims are verified, risk scores are complete and acceptable, the editorial pass has no failures, and the URL or schedule is noted. Publishing is a recorded decision, not a default.
Hold
The draft isn’t ready, and the people already involved can fix it. Holds come with a reason and a re-review date. Common causes: incomplete risk scores, a failed editorial check, or claims still marked “needs source.”
Escalate
Stop. Either the risk level is too high for you to clear alone or you’re the wrong person to decide. Escalation goes to a named person, such as the client, a subject-matter expert or an editor with authority, with a specific question.
What rules decide between them?
The Draft-to-Publish Editorial Kit uses six decision rules, short enough to memorize:
- Blank isn’t zero. If any risk score is missing, the draft holds. An unscored risk is an unknown risk.
- Any risk factor scored 2 defaults to Hold or Escalate.
- A total risk score of 5 or more holds until it’s reduced.
- Any editorial failure holds until it’s fixed.
- Claims marked “needs source” can’t be published until they’re verified, cut or escalated.
- Escalate when the topic touches health, legal, finance or safety, when the client must decide or when you lack authority.
The first rule deserves emphasis. Many scoring systems silently treat blanks as zeros, which means a reviewer who skips a question makes the draft look safer. That’s exactly backward. The kit’s workbook formulas refuse to treat blanks as zero, so an incomplete review can’t masquerade as a clean one.
What are the five risk factors?
The risk rubric scores five factors, each 0, 1 or 2:
| Factor | What a 2 looks like |
|---|---|
| Unverified factual claims | Many remain, or any critical claim is unverified |
| Topic sensitivity | Health, legal, finance or safety content |
| Invented or vague citations | Named sources you can’t open, or fake-looking journals |
| Client AI policy risk | The client bans AI and the draft’s origin is unknown |
| Brand or legal exposure | Defamation risk, trademark misuse or guaranteed-result claims |
Five factors, three levels, simple arithmetic. You can score a draft in a few minutes, and the score tells you something specific: not “this feels risky” but “this has unverified claims and a vague citation.”
Where does the claim log fit?
Before you can score risk, you need to know what the draft claims. The claim log lists each factual claim with its source, the source’s date and a status: Verified, Needs source, Cut or Escalate.
It’s tedious for about fifteen minutes and then it’s liberating, because every later decision rests on it. When someone asks why a draft was published, you can show them. When someone asks why it was held, you can show them that too.
What does the editorial pass add?
Risk isn’t the only reason to hold. The editorial pass catches the failures that appear in rushed publishing: voice drift, structural problems, a missing or confusing CTA and on-page basics like titles, headings and links. It’s deliberately focused on what goes wrong under deadline, and it pairs well with a brand book or a full SEO audit when you need either.
Any Fail in the editorial pass means Hold. Fix it, re-check and move on.
What does a decision look like in practice?
The kit includes a fictional walkthrough, Harbor & Hearth, with a before-and-after draft and annotations. In it, a draft describes a product as “healthier,” which triggers an escalation because the language is nutrition-adjacent. The decision path: hold until the nutrition language and a URL are confirmed, then publish once the editorial pass shows zero failures.
That’s the gate working as designed. One word changed the verdict, and the reason is on record.
How should you write an escalation?
An escalation is only as good as the question it asks. “Can you look at this?” sends the problem back without context. A strong escalation note has four parts:
- The specific passage, quoted, with its location in the draft.
- Why it’s escalated, in one sentence: sensitive topic, client decision, unverifiable claim or authority.
- The options, usually two or three: keep with a named source, rewrite to a narrower claim or cut.
- The deadline, tied to the publish date.
For example: “Paragraph four says our cleaner is ‘safe for pets.’ That’s a safety claim and I can’t verify it from the materials provided. Options: (a) keep, citing the manufacturer’s documentation, (b) change to ‘formulated without the ingredients listed on our safety page,’ or (c) cut. Need your call by Wednesday for a Friday publish.”
A note like that takes two minutes to write and usually gets answered in one reply. It also makes the person receiving it look good, because their decision is informed and recorded.
For freelancers, escalation is also a relationship tool. Clients notice when a writer protects them from a claim they didn’t know was risky. It’s one of the fastest ways to move from vendor to trusted partner.
How do you introduce a gate to a team that’s never had one?
Start small and make it visible:
- Use it on every piece for one month, including the easy ones. Easy pieces build the habit.
- Record verdicts in one place, so the team can see the pattern of holds and their causes.
- Agree on escalation contacts in advance. Escalation fails when nobody knows who receives it.
- Review holds weekly. Recurring causes point to upstream fixes in briefing or research.
Within a few weeks, the gate stops feeling like extra work and starts feeling like the thing that lets everyone publish confidently.
Quick answers
Does every blog post need a formal gate?
Every post benefits from the decision rules. Very low-risk posts move through in a few minutes. The gate scales with the risk rather than adding fixed overhead.
Who should make the final call?
Whoever has authority for that content. For freelancers, that may be you for Hold and the client for Escalate. Name it before you need it.
Is Hold a failure?
No. Hold is the system working. A held draft is a problem caught before a reader caught it.
Give every draft a verdict
A draft that’s published on a feeling carries risk nobody measured. A draft that’s published through a gate carries a record of what was checked and why it cleared. The difference doesn’t show on publication day. It shows the first time someone asks a hard question.
Get the Draft-to-Publish Editorial Kit: a claim-source-date log, a five-factor risk rubric that refuses to treat blanks as zero, an editorial pass and a Publish, Hold or Escalate decision sheet, with the annotated Harbor & Hearth walkthrough to show it in motion. Give your next draft a verdict instead of a vibe.
The Draft-to-Publish Editorial Kit is on Content & SEO Workflows.
Leave a Reply