Role-Task-Format Strategy Prompt
Structure AI strategy requests around expertise, objectives, and repeatable outputs
- Difficulty
- Moderate
- Time to result
- ~days to results
- Steps
- 7
- Confidence
- 99%
The Role-Task-Format Strategy Prompt uses three layers to guide an LLM toward practical marketing analysis. First, the role establishes the model's perspective by describing an experienced practitioner who has solved a comparable growth problem. Second, the task specifies the company, market, revenue stage, target, and need to validate scalable and differentiated channels. Third, the response format creates a repeated loop for each proposed channel, requiring tactics, evidence, positioning, risks, budget, necessary skills, and an implementation guide. The resulting options can be compared on common dimensions rather than presented as an unstructured list. Human expertise remains essential: the operator must add omitted variables such as average contract value and iteratively challenge questionable recommendations.
Origin
Extracted from Marketing Against The Grain, where Kieran Flanigan demonstrated a structured Claude 3 prompt for developing a SaaS marketing strategy.
Core principles
- 01AI strategy quality depends on the operator's ability to specify strategy
- 02A precise expert role supplies relevant perspective and standards
- 03The task should define business stage, objective, and validation requirements
- 04A repeated output schema makes alternatives comparable
- 05Risks, resources, budgets, and implementation steps belong in the same analysis
How to run it
- 1
Assign an expert role
Describe the kind of practitioner whose judgment the response should emulate, including relevant industry and scaling experience. Add the capabilities that matter to the task.
Pro tip Match the role's experience to the business stage and target outcome.
Watch out A prestigious but irrelevant persona will not compensate for missing context.
- 2
Specify the business context
State the product category, target customer, current scale, desired scale, and important commercial constraints. Include factors such as average contract value when they affect channel viability.
Pro tip Explain what makes the business different from a generic company in its category.
Watch out Missing economics can cause the model to recommend channels that cannot support the business.
- 3
Define the strategic task
Ask the model to identify and validate marketing channels capable of sustaining the growth target. Require differentiation from competitors and an actionable plan.
Pro tip Separate scale channels from tactics suitable only for an early startup.
Watch out Do not ask merely for marketing ideas if the real need is a scalable strategy.
- 4
Create the response schema
Define the overall strategy summary and the exact fields required for every channel. This constrains the response to decision-relevant information.
Pro tip Use clear tags or headings to distinguish role, task, and format sections.
Watch out Formatting alone cannot correct an underspecified objective.
- 5
Repeat analysis by channel
For each recommended channel, require tactics, validation, differentiation, risks, budget, skills, and implementation steps. Keep the fields consistent so channels can be compared.
Pro tip Ask the model to state assumptions behind budgets and validation claims.
Watch out A list of tactics without risks and resource requirements creates false confidence.
- 6
Critique the first output
Review recommendations using human domain expertise and identify questionable choices or missing constraints. Feed those corrections into the next iteration.
Pro tip Challenge at least one channel and ask what evidence would reverse the recommendation.
Watch out Do not mistake a detailed response for a correct strategy.
- 7
Turn the result into a reusable template
Preserve the prompt structure while making company-specific inputs replaceable. Let the team iterate on the template as new evidence emerges.
Pro tip Version the prompt alongside notes on which changes improved output.
Watch out A reusable prompt still requires refreshed business data and human review.
In the wild
The demonstrated prompt assigns Claude the role of a marketer who has scaled SaaS companies from $10 million to $100 million in ARR. It asks for differentiated, sustainable channels for an AI productivity product serving sales representatives, then requires a repeated analysis of tactics, validation, proposition, risks, budget, skills, and implementation.
→ Claude returns a detailed channel strategy covering content and search, paid acquisition, partnerships, integrations, and events, with budgets and hiring requirements.
Common mistakes
Omitting business economics
The hosts noted that missing average contract value made it harder to judge whether events and conferences were appropriate.
Accepting the first draft
The initial response is a starting point; constraints, budgets, and questionable channel selections should be challenged iteratively.
Using detail as proof
A long step-by-step answer may still rest on weak assumptions or recommend channels that do not fit the business.
Is it for you?
Best for
It is best for experienced operators using an LLM to draft or challenge growth strategies for a clearly described business.
Not ideal for
It is not ideal when the operator lacks enough market knowledge to judge assumptions, supply constraints, or identify weak strategic output.
From the transcript
“I first of all start with role, right? I give it a role.”
“Then we give it a task”
“and then I tell it the exact format”
From the episode
Mastering AI with Claude: From Dashboards to Apps