Project Management Prompts: Task Breakdown, Risk Assessment

Most project slippages are not engineering failures — they are planning failures. A vague task list, an empty risk register, and a status update that hides the truth. AI can sharpen all three, but only if you brief it like a programme manager, not a chatbot.

1. Introduction

Project management is the art of turning intentions into outcomes through other people. The mechanics — work-breakdown structures, dependency maps, risk registers, status reports — are predictable. They follow patterns that AI is genuinely good at producing. The judgement — what to prioritise, who to push, when to escalate — stays with the human PM.

This tutorial covers three of the highest-leverage PM prompts: breaking a goal into a clean WBS, building a usable risk register, and writing weekly status updates that surface problems early.

2. The Concept Explained

A project is just a tree. At the top is the outcome you want. Below it are the workstreams. Below each workstream are tasks. Below each task are subtasks small enough that one person can finish them in a day or two. Most messy projects fail because someone tried to plan two levels above where the actual work happens.

3. The Problem Without This Technique

break down a website redesign project into tasks

4. The Solution

Act as an experienced project manager who specialises in
small to mid-sized digital projects.

Project: Redesign the marketing website for a B2B SaaS
company called Tideglass (analytics for ecommerce). The
existing site has 22 pages. We are migrating from
WordPress to Webflow. Brand refresh is in scope.

Team: 1 PM (me), 1 designer, 1 front-end developer,
1 content writer, 1 SEO consultant (3 days/week).

Constraints:
- Deadline: 10 weeks from kickoff
- Budget for tools and contractors: INR 8 lakh
- Hard requirement: zero SEO traffic drop in week 1
  after launch

Produce a work-breakdown structure with three levels:

Level 1: Workstreams (4–6 of them — e.g. Discovery,
Design, Build, Content, SEO Migration, Launch).

Level 2: Tasks under each workstream. Each task must
include:
- A short verb-led name
- The single owner from the team above
- An estimated duration (in working days)
- Dependencies on other tasks (by task name)

Level 3: For the three highest-risk tasks, list two
subtasks each.

After the WBS, add a separate Risk Register table with
columns: Risk | Likelihood (L/M/H) | Impact (L/M/H) |
Owner | Mitigation. Include 6–8 realistic risks for a
project of this kind.

Rules:
- Tasks should be 1–5 working days. Anything longer
  must be broken down further.
- Every task has exactly one owner. No "shared owners".
- If something is unclear, write "[needs PM decision]"
  rather than guessing.

5. Step-by-Step Breakdown

1. State the outcome, not the activity. "Migrate the site to Webflow with zero SEO drop" beats "redo our website". The outcome shapes the entire plan.
2. Name the team and constraints. Owners and deadlines are the difference between a plan and a wish list.
3. Demand three levels: workstreams, tasks, subtasks. Anything flatter hides complexity; anything deeper is over-engineered.
4. Force single owners. "Shared owners" is the single biggest cause of late projects. Tell the AI to refuse them.
5. Always demand a risk register alongside the WBS. A plan without explicit risks is half a plan.
6. Use a second prompt for status updates. Weekly: "Here is what we did, what we are doing, what is blocked. Rate the project status as Green / Yellow / Red and justify in two sentences." Force the colour rating — it removes ambiguity.

Tip: For risk registers, ask AI to also "list two early-warning signals per risk — concrete things the team would observe if this risk were starting to materialise". Risks you can detect early are risks you can actually manage.

6. Practice Exercises

7. Key Takeaways