The two-question grid for deciding what to build
Impact and effort: a simple way to choose which AI project comes first.

Once you have a list of ideas, you need a way to choose. OpenAI’s guide uses a grid with two questions. How much would this help? How hard is it to do?
Quick wins: high impact, low effort
Do these first. Example: a review request skill that drafts a personal message after each finished job.
Big worthwhile builds: high impact, high effort
Plan these carefully. Example: a reporting view that pulls bookings, sales, ads and analytics into one place, like the connected medspa reporting we built for a client.
Self-serve: low impact, low effort
Let people do these on their own with a good prompt. Example: rewording a social caption.
Leave alone: low impact, high effort
Park these. If it is hard and barely helps, it can wait.
Two questions are enough to stop the wrong project starting first.
Score honestly
Score impact by how often the work happens and what it costs when it goes wrong, not by how exciting it sounds. Score effort by data, tools and people involved. When two items tie, choose the one your team is most annoyed by: motivation counts.
Sources
- OpenAI, “Identifying and scaling AI use cases” (PDF) — figures, examples and frameworks referenced in this article, adapted by B2 Builds for small businesses.