Planned vs actual analysis is only as useful as the analysis you actually do. These five prompts handle the most common analysis steps — paste your data, copy the prompt, and get structured insight in under a minute.
Each prompt is designed to work with data in any reasonable format. Add your actual time log where indicated.
Prompt 1: The Weekly Variance Review
Use this: Every Friday afternoon after compiling your week’s time log.
What you need: A list of your tasks for the week, with estimated time and actual time for each.
Here's my time log for this week. Each entry includes a task name, category, estimated time in minutes, and actual time in minutes.
[PASTE YOUR WEEKLY LOG HERE]
Please:
1. Calculate my overall variance rate (total actual / total estimated × 100)
2. Calculate variance rate by category
3. Identify the three task types with the largest and most consistent overruns
4. Flag any individual tasks with variance over 50%
5. Note approximately what percentage of my actual hours went to tasks not on the original plan
6. Summarize the two or three most actionable patterns you see
Be specific and direct. I want actionable insight, not a summary of the data I just gave you.Prompt 2: Pattern Detection Across Multiple Weeks
Use this: After you have three or more weeks of data accumulated.
What you need: Two or more weeks of time logs in the same format.
Here are my time logs for the past [X] weeks. Each week is labeled, and each entry has task name, category, estimated time, and actual time.
[PASTE MULTI-WEEK LOG HERE]
Please:
1. Calculate my weekly overall variance rate for each week and note whether it's trending better or worse over time
2. Identify which categories show the most consistent variance across all weeks (not just this week)
3. Note any task types where variance has been improving — these are calibrating well
4. Note any task types where variance is persistently high despite multiple weeks of data
5. Identify any time-of-week patterns if visible (e.g., Monday plans vs Thursday actuals)
6. Recommend two or three specific adjustments to my planning defaults based on this dataPrompt 3: Reference Class Estimate Generation
Use this: When you need to estimate how long a specific upcoming task will take and want to anchor to your historical data.
What you need: A description of the upcoming task, plus several weeks of historical data that includes similar tasks.
I need to estimate how long the following task will take: [DESCRIBE YOUR UPCOMING TASK]
Here is my historical time log data, which includes several similar tasks. Look for tasks in the same category or with similar descriptions.
[PASTE RELEVANT LOG DATA HERE]
Please:
1. Identify which historical tasks are most similar to what I've described
2. Calculate the average actual time for those similar tasks
3. Note the range (shortest to longest actual)
4. Give me a realistic estimate range for the upcoming task, anchored to my historical data
5. Flag if the estimate range is wide (high variance in historical data) vs narrow (reliable), so I know how much buffer to build in
6. Suggest the planning commitment I should make to my team/stakeholders given this dataPrompt 4: Deadline Feasibility Check
Use this: Before committing to a project deadline, especially when the project involves task types where you have historical variance data.
What you need: A description of the project and its major tasks, your available hours, and relevant historical variance data.
I need to evaluate whether a project deadline is realistic before committing.
Project: [DESCRIBE THE PROJECT AND ITS MAJOR COMPONENTS]
Deadline: [DATE]
My available working hours between now and the deadline: [HOURS, accounting for existing meetings and commitments]
Here is my historical planned vs actual data for the main task types involved:
[PASTE RELEVANT HISTORICAL DATA]
Please:
1. Estimate the total work required for this project, using my historical averages for each task type (not my naive estimates)
2. Apply my historical variance multipliers to each component to get a realistic total
3. Compare the realistic total to my available hours
4. Give me a clear go/no-go recommendation on the deadline with brief reasoning
5. If the deadline isn't feasible, suggest what a realistic timeline would be, or what scope would need to change to hit the stated deadlinePrompt 5: Monthly Calibration Update
Use this: Once per month to update your planning defaults based on accumulated data.
What you need: A full month of time log data with task categories.
Here is my complete time log for the past month. I want to update my planning defaults based on this data.
[PASTE FULL MONTH LOG HERE]
Please:
1. For each task category present in the data, calculate:
- Total estimated hours vs total actual hours
- Average variance multiplier (actual / estimated)
- Standard deviation of variance (to indicate how reliable the average is)
2. Identify which categories have "calibrated" — variance consistently within ±15% — and which still have significant bias
3. Generate a personal reference table in this format:
| Category | My naive estimate | Recommended planning multiplier | Notes |
4. For categories with high standard deviation (unreliable variance), explain what might be causing the variability and what I could do to make those tasks more predictable
5. Note if my overall accuracy is improving, stable, or degrading compared to what a month-one baseline might look likeGetting the Most From These Prompts
A few notes on using these effectively:
Data quality matters more than data volume. A clean log of 15 tasks with accurate actual times produces better analysis than a messy log of 40 tasks with rough guesses. Prioritize honesty over comprehensiveness.
Ask follow-up questions. These prompts generate initial analysis. The most useful insights often come from follow-up: “Why do you think my communication tasks have such high variance?” or “What would it take to get my meeting estimates to within ±15%?”
Be specific about format when needed. If you want the output as a markdown table for easy reference, say so. If you want bullet points rather than paragraphs, specify that. The prompts above leave format to the AI’s judgment; you can override that.
Repeat regularly. A single variance analysis is interesting. A pattern across weeks is actionable. A trend across months is transformative. These prompts produce their greatest value through consistent, regular use.
Related: How to Do Planned vs Actual Time Analysis — The Reality Check Loop Framework
Suggested tags: AI prompts, planned vs actual, time analysis, productivity prompts, knowledge work
Frequently Asked Questions
-
What format should I use when pasting time log data into an AI?
Any consistent format works. The simplest is a plain text table with columns for task name, category, estimated time (in minutes), and actual time (in minutes) — one task per line, with a header row. The AI will parse whatever format you give it as long as it's consistent. CSV, markdown table, or even comma-separated plain text all work. Don't worry about perfect formatting; a well-structured rough list is sufficient for variance calculations.
-
Can I use these prompts with any AI assistant?
Yes. These prompts are written for general-purpose AI assistants — Claude, ChatGPT, and similar tools. The quality of output depends on the quality of the data you provide and how clearly the prompt specifies what you want. The prompts below are designed to be specific enough to generate useful output without requiring the AI to have any prior context about your work.