≈787 tokens on demand. Your agent reads this file only when SKILL.md points to it.
Effort Estimation for Analysis Work
Guidelines for estimating how long analysis tasks will take. Use these as starting points, then adjust based on experience.
Reference Estimates by Task Type
Data Exploration and Profiling
| Task |
Estimate |
Notes |
| Run EDA on a new dataset (< 20 columns) |
1–2 hours |
Using programmatic-eda scripts |
| Profile a large table (> 50 columns) |
3–4 hours |
Include time for anomaly investigation |
| Understand a new data source (unfamiliar system) |
0.5–1 day |
Add time for documentation review and Q&A |
SQL and Data Extraction
| Task |
Estimate |
Notes |
| Simple aggregation query (1–2 tables) |
30 min |
Familiar tables |
| Complex multi-join query (3+ tables) |
2–4 hours |
Include test and validation time |
| Query a new/unfamiliar schema |
2× baseline |
Add time for schema exploration |
| Build a reusable CTE-based data model |
0.5–1 day |
|
Analysis and Modelling
| Task |
Estimate |
Notes |
| Cohort / retention analysis |
0.5–1 day |
|
| A/B test results analysis |
2–4 hours |
Assuming clean experiment data |
| Funnel analysis |
3–5 hours |
|
| Segmentation / clustering |
0.5–1 day |
|
| Predictive model (standard algorithm, clean data) |
1–3 days |
|
| Predictive model (new domain, messy data) |
3–5 days |
|
| Forecasting (time series) |
1–2 days |
|
Stakeholder Communication
| Task |
Estimate |
Notes |
| Write one-page findings summary |
1–2 hours |
|
| Build a slide deck (5–8 slides) |
3–5 hours |
|
| Build a dashboard (new, 3–5 charts) |
0.5–1 day |
|
| Write a full analysis report |
0.5–1 day |
|
| Incorporate stakeholder feedback (1 round) |
1–3 hours |
|
Estimation Adjustments
Apply multipliers to the base estimate:
| Condition |
Multiplier |
| Familiar data, familiar question type |
1.0× |
| New data source |
1.5× |
| Data quality issues suspected |
1.5–2× |
| Stakeholder hasn't confirmed requirements |
1.5× (risk of rework) |
| First time doing this type of analysis |
2× |
| Collaboration with another analyst |
0.7× (parallel work) |
| Context switch (not your primary focus) |
1.3× |
Buffer Rules
- Always add a 10–15% buffer for unexpected data issues
- For projects over 3 days: add a half-day contingency per week of work
- For any analysis with an external dependency (access request, data delivery): add 1–2 days for the dependency
Common Under-estimation Patterns
| Trap |
Why it happens |
Fix |
| Forgetting validation time |
Analysts plan for the happy path |
Add 20% of analysis time for testing and QA |
| Assuming data is ready |
It rarely is |
Check availability before committing to a date |
| Forgetting stakeholder feedback cycles |
One round of feedback = at least 2 hours |
Budget for at least one revision round |
| Underestimating unfamiliar tools |
Confidence in the method, not the tool |
Add ramp-up time for new tools |
| Not accounting for interruptions |
Focus time is rare |
Treat 6 hours as a full productive day, not 8 |
