BananaNL Blog

ChatGPT / 5 min

ChatGPT project plan prompts: separate the problem, evidence, options, and decision

A practical ChatGPT project plan prompt guide for drafting an internal decision brief by separating the problem, verified evidence, audience, options, constraints, KPIs, and open questions.

BananaNL in 1 minute

A quick look at how selected prompts move into NotebookLM and AI Chat input fields.

Watch on YouTube

A request to “write a project plan” can omit the assumptions needed for a decision

ChatGPT can help turn scattered notes into a project-plan outline, a list of discussion points, or a draft for an internal meeting. But if you provide only a topic, the result can become a generic structure without a clear problem, verified evidence, decision owner, or decision to make.

OpenAI’s prompt guidance recommends clear objectives, enough context, a specified format and tone, and iterative improvement after reviewing the output. For a project plan, separate the problem, audience, verified facts, hypotheses, options, constraints, and requested decision so an AI draft does not replace human judgment.

Practical guides to plan writing commonly put the purpose, audience, format, required information, costs, and expected outcome up front. A structure such as 6W3H can reveal missing planning details, but AI-generated figures or market statements are not evidence. Verify sources, budgets, KPIs, and delivery feasibility with the responsible people and primary materials.

An abstract illustration of a ChatGPT project-plan prompt separating the problem, evidence, options, delivery conditions, and decision

Six fields to separate in a project-plan prompt

  1. State the decision in one sentence: whether to start an initiative, change a priority, or move to the next research step. This gives the meeting a clear outcome.
  2. Separate the problem from the audience. Mark the current pain, people affected, and business impact as either verified observations or hypotheses.
  3. Set evidence rules. Name allowed sources, dates, units, and unverified items, and tell ChatGPT not to invent facts, figures, or market context.
  4. List options and constraints. Include the status quo, budget, staffing, timing, policy, and dependency constraints so the draft does not assume an impractical route.
  5. Fix the output sections: summary, background, problem, options, recommendation rationale, delivery plan, KPI, risk, open questions, and meeting decision.
  6. Keep review boundaries. Market sizing, ROI, legal questions, personal data, competitor comparisons, and budget approval need source checks and human ownership.

Four inputs to prepare for a project-plan draft

The decision requested in this meeting, intended reader, purpose, and formatVerified current state, problem, source highlights, source dates, and evidence linksAudience, options, deadline, and budget, staffing, or operating constraintsHow success will be measured, KPI candidates, expected risks, and unresolved questions

How to shift the brief for internal project plans

New-initiative proposalSeparate the problem, audience, hypothesis, delivery approach, and resources. Treat expected effects as measurement proposals, not confirmed outcomes.
Process-improvement planProvide the current workflow, bottleneck, affected scope, migration conditions, and exceptions. Leave efficiency figures as items that require measurement and team confirmation.
New feature or service discoveryOrganize user problems, verified usage evidence, alternatives, technical and operating constraints, and the smallest test. Review requirements and implementation decisions separately.
Leadership or department meeting briefLead with the requested decision, options, trade-offs, principal risks, and next deadline. Keep the background concise and preserve references to evidence.

ChatGPT project-plan prompts to try

Use 6W3H to expose gaps without inventing evidence

Organize the project notes below with 6W3H for an internal planning review. Do not add facts, figures, market trends, or customer feedback that are not in the input. Mark every point as Verified, Hypothesis, or Open. List missing information as questions. Then create a project-plan outline in this order: requested decision, background, problem, audience, options, delivery conditions, measurement, risks, and next checks. Notes: """[paste notes here]"""

It keeps an early idea from looking like a finished plan and makes evidence gaps visible.

Create a decision-meeting outline

Use only the verified information below to create a project-plan outline for [meeting audience] to decide [decision requested]. Format: [slides / document]. Length: [slide count or word count]. For every section, include: conclusion, evidence used, unresolved items, and the decision needed in the meeting. Include the status quo among the options and list benefits, constraints, and needed follow-up for each. Do not estimate effects or costs; label unknown figures ‘needs confirmation.’ Information: """[paste background, problem, audience, options, constraints, and source highlights]"""

It produces a consistent decision brief rather than a chapter list with unsupported conclusions.

Review a draft before approval

Review the project-plan draft below before approval. Before rewriting it, classify statements as: 1) verified facts, 2) hypotheses or assumptions, 3) claims that need a source, 4) items needing budget, legal, privacy, or operating review, and 5) unclear decision requests. Do not make legal or business-viability conclusions. Briefly explain why each item needs checking and suggest a responsible review role. Draft: """[paste draft here]"""

It finds missing decision conditions and review ownership before the document is polished.

Save reusable project-plan prompts in BananaNL

Project plans repeatedly need the same fields: requested decision, problem, evidence, options, constraints, KPIs, risks, and open questions. Save templates for initial planning, decision briefs, and review in BananaNL, then replace only the project-specific primary information.

BananaNL is a Chrome extension that inserts saved prompts into the input fields of AI Chat services such as ChatGPT, Gemini, and Grok. It never auto-sends, so you review source material, figures, approval scope, and publication status before sending yourself. Under the current pricing boundary, NotebookLM can be started for free. Viewing, inserting, and saving AI Chat prompts are paid features, while Image Collections, image saving, image editing, and video conversion are free features.

An abstract illustration of retrieving a saved project-plan prompt template from BananaNL

FAQ

Can ChatGPT write a project plan from scratch?

It can help with a structure or discussion draft, but provide verified problems, figures, sources, budgets, and delivery conditions. Do not treat AI-added material as evidence; the responsible team makes the decision and approval.

How is a project-plan prompt different from a requirements prompt?

A project plan organizes why to act, which option to choose, and what decision is requested. Requirements work then details the scope, functions, conditions, and acceptance criteria for the approved direction.

Can I ask for KPI or cost figures?

Use ChatGPT to organize candidates and review questions, not as a substitute for actuals, budgets, quotes, or market data. Check each figure’s source, unit, period, and calculation method.

If searching for prompts is the hard part, use BananaNL

Prompts become useful when they are close to the input field. Use BananaNL to carry them there, then adjust before sending.