How to Decide Whether a Client Project Should Be Hourly, Fixed Fee, or Retainer With AI

The problem and who this is for

This workflow is for freelancers and consultants who know how to price work but still choose the wrong pricing model too often. The most common mistake is using a fixed fee when the scope is not stable. The second most common mistake is staying hourly when the deliverable is actually clean enough to package.

You can avoid both errors by reviewing the request against a few decision factors before you build the quote.

Prerequisites

You need the client request, your discovery notes, or both. You also need a short description of how you usually deliver the work. A project that looks fixed-fee friendly on paper may still be bad for fixed pricing if your own process depends on many external approvals.

ChatGPT is the best fit because this is a reasoning workflow built mostly from pasted text. Gemini and Claude are both practical alternatives.

How to gather the source material

1. Copy the client's request into a document or note.
2. Add your discovery notes below it.
3. Write one line each for these four factors:
   • scope clarity
   • revision risk
   • cadence
   • dependency load
4. If the client is asking for ongoing access, review, or recurring changes, note that separately. That is often the strongest retainer signal.

The workflow

1. Paste the request and your notes into ChatGPT.
2. Ask it to compare hourly, fixed fee, and retainer pricing for this exact scenario.
3. Read the explanation, not just the recommendation.
4. If it recommends fixed fee, ask what must be clarified before that model is safe.
5. If it recommends hourly, ask whether a small fixed-fee discovery phase should come first.
6. If it recommends retainer, ask what recurring deliverable or cadence would justify it.
7. Build the actual quote only after the pricing model makes sense.

Primary tool instructions: ChatGPT

1. Use one new chat per project so the reasoning stays clean.
2. Give the tool structured inputs, not a long emotional rant about the client.
3. Ask for a recommendation and a backup option.
4. Make it explain why the rejected models are weaker. That is often where the real insight lives.

Alternative tool instructions

Gemini

Gemini is a good fallback if the request already lives in a Google Doc or email export. Paste the same structured notes and use the fallback prompt.

Claude

Claude is a good alternative when the project request is messy and you want a more careful written explanation of tradeoffs before you choose a model.

Copy and paste prompt blocks

Primary prompt for ChatGPT

{
  "role": "pricing-model-selector",
  "goal": "Choose the best pricing model for a new client request.",
  "inputs": {
    "project_summary": "Paste the request or discovery notes.",
    "scope_clarity": "Describe how well-defined the work is.",
    "revision_risk": "Low, medium, or high.",
    "cadence": "One-time, phased, or ongoing.",
    "dependency_load": "List approvals, client inputs, and outside dependencies.",
    "my_delivery_pattern": "Describe how I usually work."
  },
  "instructions": [
    "Compare hourly, fixed fee, and retainer pricing for this specific request.",
    "Explain which model protects scope and margin best.",
    "Flag what would have to be true for a fixed fee to be safe.",
    "End with a recommendation and a backup option."
  ],
  "output_format": {
    "sections": [
      "Project Read",
      "Hourly Fit",
      "Fixed Fee Fit",
      "Retainer Fit",
      "Best Choice",
      "Backup Choice",
      "Why"
    ]
  }
}

Fallback prompt for Gemini or Claude

{
  "role": "pricing-model-reviewer",
  "goal": "Pressure-test which billing model fits a project.",
  "inputs": {
    "request_text": "Paste the request.",
    "unknowns": "Paste the biggest scope unknowns.",
    "my_constraints": "Optional"
  },
  "instructions": [
    "Do not answer in generalities.",
    "Use the actual project details to explain the recommendation.",
    "Show what makes each model risky or safe."
  ],
  "output_format": {
    "sections": [
      "Key Unknowns",
      "Safest Model",
      "Why",
      "Conditions For Other Models"
    ]
  }
}

Quality checks

• The recommendation should point to actual risk factors, not generic best practices.
• If the tool recommends fixed fee, it should name the assumptions that make that safe.
• If the work is ongoing or cadence-based, retainer should get real consideration.
• If the work is open-ended, hourly should stay in play even if fixed fee sounds nicer to the client.

Common failure modes and fixes

The tool picks fixed fee too often

Add more detail about revision risk and dependency load. Those two variables usually correct the bias.

The tool stays too abstract

Force it to compare the three models side by side with explicit risks.

The client wants fixed fee but the project is still fuzzy

Ask the tool to draft a small paid discovery phase first.

You already know the answer but want confirmation

That is fine. Use the workflow as a discipline check before pricing, not as a replacement for judgment.

Sources Checked

• https://help.openai.com/en/articles/10169521-projects-in-chatgpt (accessed 2026-03-24)
• https://support.google.com/gemini/answer/14903178 (accessed 2026-03-24)
• https://support.anthropic.com/en/articles/9517075-what-are-projects (accessed 2026-03-24)

Quarterly Refresh Flag

Review by 2026-06-22. Recheck project workspace behavior and any material changes that affect long-running pricing decision chats.