The first question in every content workshop is “what should we write about”, and it is the wrong one. The topic is usually decided by the product, the season or the search demand. What nobody has decided is the content angle, and that is the decision that decides whether anyone reads the piece.
Code: AngleCraft, MIT licensed.
I teach the angle as a separate step with its own exercise, because writers skip it when it is folded into the brief. Here is the version that survives contact with a real team.
The angle is a decision, not a style
Same building, ten photographs. The facade sells the address, the rooftop sells the view, the reflection in a puddle sells the mood. Nothing about the building changed.
For a topic, the lenses look like this: a myth worth disproving, a paradox worth resolving, a real cause behind a visible symptom, the true cost of a solution that works, the branches of a decision, the driver nobody names, a contrarian claim that has to defend itself, the history, the forecast, the news that makes it urgent today.
Seven of those are perspectives. Three are wider lenses, and I keep them separate because a lens covers many topics where a perspective covers one claim.
The exercise I give writers
Take the topic and write one hook for each of the ten lenses. One sentence, thirty words maximum, no research, no drafting.
Then keep three. Not the three you like, the three a colleague would stop scrolling for. The rest go into a file, because a discarded hook is a future article and not a failure.
The exercise takes forty minutes and it changes what the writer researches. A hook built on a real cause needs a different interview than a hook built on a forecast, and knowing that before research saves the whole week.
Three checks I run in review
Surprise: does the hook contradict something the reader believes? If it confirms what they already think, it is a summary, not an angle.
Specificity: is it anchored in a mechanism, a trade-off or a rule the reader can apply? “Caching speeds up your site” fails; “the cache key decides whether a deploy invalidates the page” passes.
Curiosity gap: does the reader need the next sentence? A hook that answers its own question has nothing left to sell.
A hook that fails specificity cannot be rescued later. The reader decides in the first line, and no amount of polished paragraphs brings them back.
The placeholder rule
If a hook needs a number and nobody has that number, write [N] and stop. An invented statistic is the fastest way to lose the one reader who checks.
That rule matters more with AI in the loop, not less. A model asked for a hook will supply a plausible percentage without hesitating. The placeholder makes the missing fact visible, and visible gaps get filled; invisible ones get published.
Where the lenses fail
Conditions need real branches. “If you are a solo founder” is not a branch when the audience is all solo founders.
The news lens expires within a week, and a piece published on day nine reads as old news with a lesson attached.
The contrary thesis needs an honest admission of where the opposite stops being true. Without it the piece is contrarian for the sake of the format, and readers notice.
What I ask teams to stop doing
Writing the outline first. An outline is the article, and by then the angle is fixed by accident.
Publishing the same lens ten times. Ten posts from the myth angle is one post, repeated, and the audience learns that the feed is predictable.
Asking for the full draft at the angle stage. The angle is a decision about attention; a 1 500-word draft at that point is a commitment made before the thinking.
I use a related discipline in an AI copywriting skill built on cognitive biases: the lever is chosen deliberately, not inherited from the first draft. And before a launch I want the failure modes on the table, which is what the premortem skill is for.