A prompting technique that asks an AI model to work through a problem step by step before answering. For agencies, it improves the quality of analysis, planning and brief interpretation.
Also known as CoT prompting, step-by-step prompting
Chain of thought prompting means instructing a model to show its intermediate reasoning, for example “think through this step by step,” or giving an example that includes the working. The model then writes out the steps and arrives at the answer from them rather than jumping straight to a conclusion.
It matters most on tasks with several parts, such as comparing options, doing arithmetic or applying rules to a messy situation. Newer reasoning models do some of this internally, but the technique still helps on standard models and makes the answer easier to check.
Most agency work that goes wrong with AI goes wrong in the middle of a multi-step task, which is where this technique helps.
Briefs get read properly. Asking the model to list the objective, audience, constraints and deliverables before drafting catches misreadings early.
Budget and media math gets checked. Step-by-step working lets a planner spot the wrong assumption instead of trusting a bare number.
Reviews get faster. When the reasoning is visible, a senior reviewer can correct one step rather than rewriting the whole output.
A media planner asks a model to split a $120,000 quarterly budget across four channels given reach targets and a minimum spend per channel. The first prompt returns a tidy table that breaks the minimum on one channel. The second prompt asks for the constraints to be listed first, then each channel to be allocated in turn, with a final check against every rule. The visible steps show the error in seconds, and the planner keeps that structure as a saved template for every future split.
The automations and agents module of the workshop teaches you how to build AI workflows that compress the busywork without taking the craft out of the studio.