A
?
AssetsWorkflowsHandbookDesign system
← Handbook

SOP — Draft angles

Who: Creative Strategist · When: angle material exists and the client needs new creative Time: 10–15 minutes per batch, plus founder review

What this is for

An angle is a reason someone should choose this client, expressed as a hook a scroll-stopping ad can lead with. It is not a headline — it is the argument the headline serves. Each angle generates multiple creatives, so the leverage is here, not downstream.

The system generates angle candidates from the approved angle material. This step reviews them, kills the weak ones, and sends the survivors to the founder for approval.

Steps

  1. Run the strategy engine from the client's Creative tab. The system generates candidate angles from the approved angle material and existing insights.
  2. Review each candidate. The system shows the hook, the mechanism (why it works), and the funnel stage it targets. Read the mechanism — if it does not name a specific reason this argument would stop someone, the angle is filler.
  3. Kill anything that: - Repeats an angle already running. Diversity is structural, not a preference — the same argument twice earns half the learning. - Could be said about any company in the category. "Quality workmanship" is not an angle. - Contradicts the knowledge base. The system checks text claims but not imagery associations. - Targets a funnel stage the client already wins at. An angle solving a problem the client does not have wastes a creative slot.
  4. Approve the rest as drafts. They move to the founder's review queue.
  5. Write a one-line note on anything you're unsure about. The founder sees the note alongside the angle and can make the call without a meeting.

Judgment calls that stay yours

How many is enough. Three to five per cycle is the range. Fewer than three gives the media buyer nothing to test; more than eight cannot be tested at the budget most Atlas clients run.

Whether an insight is strong enough to build on. The system flags insights whose supporting facts have been retired, but a technically supported insight can still be weak. If the "why not generic" field does not convince you, it will not convince an audience.

What good looks like

A batch of angles where each one names a different reason to choose the client, each targets a different part of the funnel or a different audience pain, and none of them could be swapped to a competitor without rewriting the core claim.