Everyone on the content team
Led by the AI head of content marketing, under your specialist. One argument per client, written down, with everything laddering to it. Reads the business, the buyer and the field before a brief exists, and measures tone rather than asserting it.
Where this team overlaps another: The content desk owns the argument, the angles nobody took, distribution and the week-eight verdict. It does not own the keyword-led production queue, which belongs to search. When both want the same page, the point of view wins on what it says and search wins on where it sits and how it is marked up. Neither commissions a writer the other does not know about, because there is only one writer.
Establishes who the content is for and what they are trying to get done
NeverEvidence for an audience claim comes from queries, analytics or the client's own pages.Reports what the content actually did, including when the answer is nothing
NeverReport flat as flat. A quarter with no movement is a finding.Finds out what the business actually is before anybody writes about it
NeverQuote the page. A claim in your report should be findable on the site.Builds content out of original data nobody else has
NeverState the sample size and the method beside every finding, always.Decides where a piece goes after it is published, which is most of whether it works
NeverPlan distribution before the piece is written, not after it underperforms.Decides the one thing this company is arguing, and makes everything ladder to it
NeverThe point of view must be arguable. If nobody could disagree, it says nothing.Reads the pages you are competing with and measures how they are built
NeverRead the pages. A SERP listing is not a competitor analysis.Judges whether a draft is worth a reader's time, which is a different question from whether it is correct
NeverYour question is whether anyone would finish it, not whether it is accurate.Measures how a company writes and says, with evidence, whether it should change
NeverEvery tone claim carries the client's number and the comparison number.Finds the way into a topic that nobody else took
NeverRead what already exists before claiming an angle is unused.Generates deliberately unreasonable ideas so the reasonable ones have competition
NeverQuantity first. Twenty concepts, then filtering, never the reverse.Writes the title and the first three sentences, which decide whether the rest is read
NeverWrite ten titles before choosing one. The first is never the best.Turns one piece of research into everything it should have been
NeverEach derived asset must stand alone. A teaser that needs the original is not an asset.What this team will never do
- No tone verdict on fewer than three readable rival pages. Two is one writer's habit, and asking you to rewrite a site against it would be the expensive kind of wrong.
- No voice ratios under 120 words. Below that the numbers are arithmetic rather than evidence, and the screen says so.
- No point of view nobody could argue with. A thesis with no opposing position is a description.
- No claim in a draft that is not in the fact ledger with the page it came from.
Everyone on the social team
Led by the AI head of social, under your specialist. The owned channels, worked from the argument the content desk already had approved. Reads what measurably worked for the competitors it can read, says plainly which platforms publish nothing, and drafts everything for a person to approve.
Where this team overlaps another: Social is downstream of the content desk rather than beside it. It works from the point of view already approved instead of inventing a second brand voice, and the four assets inside a finished piece are pulled once by the repurposer rather than twice. Where social research turns up demand nobody has satisfied, it goes to the content desk as a subject rather than being written twice.
Reads what the audience actually asks in comments and turns it into subjects
NeverA question counts when more than one person asks it.Drafts replies to comments and messages, and escalates the ones a person must answer
NeverNothing is sent. Every reply is drafted and a person approves it.Chooses reel, carousel, static or long form, from what worked rather than from habit
NeverFour posts in a format is the floor before that format gets a verdict.Writes the first line and the first frame, which decide whether anything else is read
NeverTen hooks before one is chosen. The first is never the best one.Decides which platforms deserve the client's effort and which do not
NeverCompare engagement rate against each platform's own audience, never raw totals.Turns the measured cadence into a calendar the client can actually sustain
NeverCadence comes from the field's measured frequency, not from a round number.Writes the caption or the script, in the client's approved voice
NeverEvery factual claim comes from the fact ledger with its source.Reads competitors' public posts and reports what measurably worked
NeverTwelve posts is the floor for a median. Under it, report the floor rather than a number.The gate. Asks whether anybody would stop scrolling, which the other checks do not
NeverOrdinary is a failure state. A post nobody would stop for is a cost, not a neutral.Measures what the social work actually did, and names what did not work
NeverCompare against the client's own baseline, not against a category average.Finds what is moving in the client's category right now, from readable platforms only
NeverA trend needs the same movement across at least three accounts, not one.Specifies what the asset has to be, so a person or a tool can make it
NeverEvery brief names the first frame or the focal point explicitly.What this team will never do