Lesson 3 of 7 in Create Work That Fits the Strategy
- 1Move from a plan task to Create
- 2Start a New Asset Run and understand the cost boundary
- 3Review concepts before rendering visuals
- 4Use brand voice, typography, logos, and reusable assets
- 5Review, vote, reject, and explain a creative direction
- 6Refine in the design editor
- 7Save, keep the content-task connection, and ship
Why this matters
Ideas and visual directions are inexpensive to change; rendered images are not. This review is the last point in Execute where a weak angle or an unsupported claim costs only a few seconds to cut. It also protects trust: an image carrying an invented number or an off-strategy message costs more than credits, because once it ships it speaks for the business. Human judgment decides what goes forward; the recommended direction informs that judgment and does not replace it.
When to use it
After every idea run, before you click the image button. The same check applies to ideas you added yourself with "Add Manually".
Before you start
- A finished idea run showing under "Review Ideas" (Start a New Asset Run and understand the cost boundary).
- The plan task's audience and objective in mind, so you judge each idea against the job rather than against taste.
- Knowledge of what the business can prove: the sources behind any number, testimonial, or promise an idea might use.
Do this
- 1Under "Review Ideas", expand each idea with the arrow and read its "Text Overlay", "Caption", and "Image Prompt". Edit the text in place, or remove the idea.
- 2Test each idea against the task: right audience, right objective, right channel. Cut ideas that are clever but off the brief.
- 3Check every claim. Each number, testimonial, or promise must trace to evidence you hold (Why evidence and provenance matter); soften or cut the rest. The idea generator is told to avoid manufactured urgency in healthcare and mental-health messaging and competitor comparisons in healthcare and real estate. Do not write them back in by hand.
- 4Read the "Visual direction" under each idea. "Recommended" marks the system's pick; "Details" shows the composition, mood, what the concept "Refuses", "Why it works", and "Why recommended". Choose another concept with the buttons beside it if it fits the task better.
- 5If none fits, open "New directions", add an optional steer, and click "Try again"; the button shows its credit cost. Work one idea at a time.
- 6When every surviving idea has a direction you would defend, click the image button, which now shows how many images it will render and their credits.
What you should see
Each remaining idea shows text you agree with and one chosen visual direction. Ideas you cut are gone before any image credits are spent. If the image button still reads "Plan visual directions first", at least one idea has no direction yet.
How to judge it
For every surviving idea you can name the task it serves, the evidence behind each claim, and why you chose its direction over the alternatives. If you cannot explain why an idea survived, cut it.
What it does not mean
- "Recommended" is the system's best reading of your brand and brief, not the right answer. Your judgment overrules it.
- Cutting an idea is not wasted work. A rejected concept costs far less than a rendered image nobody uses.
- Persuasion is not invention. A punchy concept still needs true claims behind it.
Try it in Signal & Science
Open Create and review your latest run's concepts and claims before any visual renders.
Open in Signal & Science