Study Guide: Alex for Data Analysts
Your reference for using Alex across the data analysis lifecycle. Ready-to-run prompts for exploration, interpretation, storytelling, and communication.
What This Guide Is Not
This is not a habit formation guide (see Self-Study Guide for that). This is a data analysis toolkit — the specific ways Alex can accelerate your work from raw data to decisions.
Core Principle for Data Analysts
Data analysis is about turning numbers into understanding. Alex can’t run your queries or see your data — but it can help you think more rigorously about what the data means, communicate it more clearly, and structure your analysis before you open a single spreadsheet.
The key pattern: paste your actual data or output. Describe your dataset, share summary statistics, paste query results. The more concrete signal you give, the sharper the insight.
The Seven Use Cases
1. Exploratory Data Analysis Planning
When to use: Starting a new analysis. Figuring out what to look for before diving in.
Prompt pattern:
Help me plan an exploratory analysis:
Dataset: [describe the data — tables, fields, time range]
Business question: [what you're trying to answer]
What we know: [any hypotheses or context]
Stakeholder: [who needs the answer]
Help me:
1. Define the key questions to investigate
2. Identify the metrics that matter
3. Suggest the analytical approach (cohort, trend, segmentation, etc.)
4. Flag potential data quality issues to check
5. Outline the analysis in logical order
Follow-up prompts:
I found [X]. What should I investigate next?
This looks like an outlier. What are the possible explanations?
What would prove our hypothesis wrong?
2. SQL and Query Design
When to use: Designing queries, troubleshooting logic, reviewing query structure.
Prompt pattern:
Help me design a query:
Goal: [what I need to calculate]
Tables: [schema or description of relevant tables]
Filters: [what to include/exclude]
Granularity: [per user / per day / per cohort]
Special considerations: [nulls, duplicates, time zones]
Write a SQL query that:
1. Is logically correct
2. Handles edge cases explicitly
3. Is readable and commented
4. Is efficient for the likely data volume
Follow-up prompts:
This query is slow. What might be causing it?
I'm getting unexpected NULLs. What would cause that?
Refactor this to use a CTE for readability.
[paste current query]
3. Data Interpretation and Insight Generation
When to use: Making sense of results. Finding the story in the numbers.
Prompt pattern:
Help me interpret this data:
Context: [what analysis this is from, what the business question was]
Results:
[paste or describe the data]
Help me:
1. Identify the most important finding
2. Distinguish correlation from causation
3. Generate 3-5 hypotheses that could explain this pattern
4. Identify what additional data would confirm or refute
5. Surface what surprises me vs. what's expected
Follow-up prompts:
The metric went up 20% last month. What are all the possible explanations?
I think this is caused by [X]. What would prove/disprove that?
What am I not seeing in this data?
4. Data Storytelling and Visualization
When to use: Presenting findings. Choosing the right chart. Building a narrative.
Prompt pattern:
Help me tell this data story:
Audience: [who this is for — executives, technical team, external]
Key finding: [the main thing I want them to understand]
Data:
[describe or paste the key numbers]
Decision: [what they need to decide based on this]
Help me:
1. Structure the narrative arc
2. Choose the right chart type for each finding
3. Write the headline insight for each slide/section
4. Anticipate their questions
5. Make the recommendation clear
Follow-up prompts:
They'll ask "so what do we do about this?" Prepare my answer.
Write the executive summary — 3 bullet points max.
This chart is confusing. How do I simplify it?
5. Metric Definition and Measurement Design
When to use: Defining KPIs, designing experiments, setting up tracking.
Prompt pattern:
Help me define this metric:
What we want to measure: [the business objective]
Why it matters: [how it connects to outcomes]
Current proxy: [what we're measuring now if anything]
Help me:
1. Define the metric precisely (numerator, denominator, time window)
2. Identify pitfalls and edge cases
3. Find potential gaming or unintended consequences
4. Suggest leading vs. lagging variants
5. Design how we'd know if the metric is working
Follow-up prompts:
This metric goes up when [bad behavior]. How do I adjust it?
We have 5 competing metrics. Which should be the north star?
Write the metric definition for the data dictionary.
6. A/B Test Design and Analysis
When to use: Designing experiments, interpreting results, communicating findings.
Prompt pattern:
Help me with this experiment:
Phase: [design / results interpretation / communication]
What we're testing: [hypothesis]
Metric: [primary metric]
Results (if analysis phase):
[paste results]
For design:
1. Is the hypothesis testable?
2. What sample size do we need?
3. What are the risks of false positives/negatives?
4. What would make this test inconclusive?
For analysis:
1. Are results statistically significant?
2. What's the practical significance?
3. Are there segments that behave differently?
4. What's the recommended decision?
Follow-up prompts:
Results are statistically significant but the effect is tiny. What do I recommend?
The test ran for 2 weeks. Is that long enough?
Write the experiment results summary for stakeholders.
7. Stakeholder Communication
When to use: Translating technical findings for non-technical audiences. Presenting to leadership.
Prompt pattern:
Help me communicate this analysis:
Audience: [technical / business / executive / mixed]
Finding: [what I discovered]
Why it matters: [business impact]
Recommendation: [what I suggest]
Questions they'll ask: [likely pushback]
Create a communication that:
1. Opens with the finding and its business impact
2. Explains the methodology at the right level
3. Makes the recommendation clear
4. Handles the likely counter-questions
5. Calls out caveats and limitations honestly
Follow-up prompts:
They'll ask "how confident are you in this?" Prepare my answer.
Convert this to a one-page summary.
What assumptions am I making that I should state explicitly?
Practice Progression
Week 1: Plan your next analysis using the exploration prompts before starting in SQL.
Week 2: Interpret a recent dataset with the interpretation prompts. Compare to what you concluded on your own.
Week 3: Build a stakeholder presentation using the storytelling framework.
Week 4: Design or review an experiment using the A/B testing prompts.
What Great Looks Like
After consistent use, you should notice:
- More structured analysis before you start coding
- Richer interpretations with more hypotheses considered
- Clearer, more persuasive stakeholder communications
- Better experiment design with fewer surprises
The goal isn’t for Alex to analyze — it’s for Alex to help you think more rigorously before, during, and after your analysis.