The Copy Playbook

Positioning · the context that makes your value obvious

Upstream

This sits above Message Architecture. Message is downstream of positioning: what you say only makes sense once you've decided what the thing is, what it beats, and who it's for. A reader who can't place your product will judge it against the wrong yardstick, and no amount of good writing fixes that. Decide the context here, then go write.

Source The component set — and the insistence that its order is load-bearing — is adapted from April Dunford's positioning work, with the status-quo argument from her book on pitching. The names are not hers. A playbook whose engine is vocabulary creation has to name its own parts, so every component below is renamed into our language. The worked example is the live Writer campaign.

The chain — order is the content

Six questions, and they only work in sequence. Each one hangs off the answer above it: nothing is "only yours" until you've named what you're compared to, nothing is valuable until it's only yours, and no shelf is right until you know who cares. Most bad positioning is a right answer given in the wrong order.

The Instead

What would they do if you didn't exist?

The Only

What do you have that none of those have?

The So-What

What does that make possible for them?

The Tell

Who has the signal that says they'll care a lot?

The Shelf

What category makes that value obvious on sight?

The Clock · optional

What makes this matter now rather than eventually?

The chain

Six links, in order

Instead → Only → So-What → Tell → Shelf → Clock.

Each link is defined by the one above it. Answer them out of order and the answers won't hold.

Failure mode: starting at The Only (a feature list) or The Tell (a segment). Both produce positioning that reads fine internally and lands nowhere.

Yours: these names are the point. Borrowed mechanics, own vocabulary — the same move the whole system runs on.

The Instead — including doing nothing

Not "who are your competitors" — what would this person actually do if you vanished tomorrow. Usually that's a messy list: a rival tool, a manual workaround, a person they'd pay, and the quiet favourite, nothing at all. Get this list wrong and every answer below it is measured against the wrong yardstick.

What would a Writer reader do instead?

A rival toolA generic AI writer that returns a finished wall of text in a voice that isn't theirs.

A workaroundA blank doc in one tab, a chat window in another, copy-pasting between them all afternoon.

A personHire an editor, or corner a friend who writes and ask them to read it.

Nothing — the favouriteKeep the idea in their head. The draft stays unstarted. Nobody gets blamed for that.

The Instead

What they'd do without you

Products, workarounds, people, and nothing.

Somewhere between 40% and 60% of considered decisions end in no decision at all.

And usually not because they prefer the status quo — because they don't feel confident enough to choose. Doing nothing is the low-risk option.

Yours: this rewrites the Stakes job downstream. The thing to beat is inertia, not a rival.

The Only — differentiators, not table stakes

Now, and only now, list what you have that the alternatives don't. The discipline is subtraction: most of what a team is proud of is table stakes. Table stakes still have to work, but they earn no room in the copy, because everyone on the Instead list can claim them too.

✓ Only — no alternative has it

The draft and the collaborator in one place.

Scoped edits — the rest of the draft is untouched.

Research that reports where it's thin.

Your own skills, called by name.

✕ Table stakes — everyone has it

Writes fast.

Understands plain-language instructions.

Works in the browser.

Saves automatically.

The Only

What no alternative offers

Unique is a relationship, not a property.

Test each one against the Instead list. If any alternative could claim it, it's table stakes.

Table stakes must exist. They just don't get argued about.

The So-What — what the Only makes possible

An attribute isn't a benefit. Each Only has to map to something it lets the reader do, said in their terms, or it stays a spec sheet. Run the mapping explicitly — one line each — and the ones that map to nothing quietly reveal themselves.

Draft and collaborator in one place→You stop losing the thread between tabs

Scoped edits, rest untouched→You stay the author of every line

Research that flags its thin spots→You know which claims still need checking

Your own skills, called by name→It learns your craft instead of imposing one

The So-What

Attribute → what it enables

One line each. No orphans.

An Only that maps to nothing isn't a differentiator, it's trivia.

This feeds the Promise and Proof jobs downstream. Don't write them twice.

Yours: claim only what the attribute actually enables. The proof ladder still applies.

The Tell — who cares a lot, and how you'd know

Not a demographic. A tell is something you could actually observe that predicts this person will care intensely about the So-What. Job titles and company sizes are weak tells. Behaviours and existing frustrations are strong ones, because they're evidence the value already matters.

✓ Strong tells — observable

Has a folder of drafts that stop after two paragraphs.

Tried an AI writer and could hear it wasn't their voice.

Publishes on a schedule they keep missing.

Edits their own sentences out loud.

✕ Weak tells — just description

Knowledge workers.

Aged 25–45.

Interested in productivity.

Uses a computer for their job.

The Tell

The signal, not the segment

What would you literally see this person doing?

Strong tells are behaviours and frustrations. Weak tells are categories.

The tell also tells you where to find them — which makes it a distribution answer, not just a copy one.

The Shelf — the category you're judged on

People use what they already know to work out what they don't. The shelf you sit on tells them instantly what a thing is — and hands them a yardstick. Choose wrong and your best, most honest difference reads as a missing feature. This is the most consequential decision on the page, and it's the one most often made by accident.

✕ Wrong shelf

AI writing tools

Neighbours: generators, rewriters, one-click blog posts.

Implied promise: it writes it for you. Judged on output volume. "You stay the author" now reads as doing less than the shelf next to it.

✓ Right shelf

Writing collaborators

Neighbours: the editor, the co-author, the person in the room.

Implied promise: it helps you write yours. Judged on whether the work stays yours. Now the honest difference is the whole point.

The Shelf

Context makes value legible

The category hands the reader a yardstick. Pick the one your Only wins on.

The neighbours on the shelf are your comparison set, whether you named them or not.

Test: on this shelf, does my Only read as an advantage or as a gap?

Yours: this is the vocabulary fight at product level. Name the shelf and you set the terms everyone else argues inside.

The Clock — why now, and the trap in it

A trend answers why this matters right now. A shelf answers what it is. They get confused constantly, and the confusion is expensive: a trend used as a category tells the reader nothing they can act on. In our corner of the world this is the easiest mistake available, so it gets said plainly — AI is a clock, never a shelf.

✓ Shelf + clock

A writing collaborator — now that agents can hold your context and run your own skills.

The shelf says what it is. The clock says why it's newly possible.

✕ Clock alone

An AI-powered platform for writing.

Tells the reader nothing about what it does, who it's for, or what it replaces.

The rule

The shelf says what it is. The clock says why now.

The Clock

Optional, and layered on

A trend rides on top of a category. It never substitutes for one.

Trap: "AI" is a trend. Used as a category it says nothing — every product in the market can claim it.

Skip this link entirely if no honest trend applies. A missing clock is fine. A fake one is not.

The capture — and the handoff

Positioning that lives in your head isn't positioning, it's a hunch. Six lines, written down, is the artifact — and it's what Message Architecture opens before it fills in a single one of its own jobs. The two pages don't repeat each other; this one feeds that one.

Writer — positioning, captured

InsteadThe blank page and not writing it. Generic AI that hands back a stranger's voice. Two tabs and a clipboard.

OnlyDraft and collaborator in one place. Scoped edits. Research that admits its thin spots. Your own skills.

So-whatYou stop facing the blank page alone, and every line is still yours.

TellAlready writes. Has the folder of two-paragraph drafts. Can hear when a sentence isn't theirs.

ShelfA writing collaborator — not an AI writing tool.

ClockAgents that hold context and run your skills make "collaborator" real rather than aspirational.

feedsReader ← the Tell, in their own words

feedsPromise ← the So-What, said as an outcome

feedsStakes ← the Instead, specifically the doing-nothing line

The capture

Six lines, then brief

One document, shared, that the whole system inherits.

Message Architecture starts by reading this. Its old Position job now points here.

Can't fill a link? That's research, not writing. Close it before you draft.

Yours: revisit when the Instead list changes. New alternatives move every link below them.

Created for you, with ❤️ by Robert Evans · The AI Design Architect