A portfolio after the Google Data Analytics Certificate should make your work understandable and verifiable. Begin with one complete case study, then add a project that demonstrates a different capability needed for your target roles. There is no universal three-project or five-project minimum that guarantees interviews. Depth, relevance, and clarity matter more than collecting dashboard screenshots.
The Google capstone course includes using case studies and portfolios when communicating with potential employers. Your portfolio can extend that work, but a completed course exercise is not professional employment. Label guided work, independent work, and any team contribution accurately.
What should the first projects demonstrate?
Choose projects by missing evidence, not by tool count. A capstone might show the full Ask, Prepare, Process, Analyze, Share, Act sequence. A second project could focus on a difficult SQL join or a spreadsheet reconciliation. A third might show a dashboard designed for a specific audience. Building the same sales dashboard in three tools provides less new evidence than solving three different problems.
| Artifact | Question it answers for a reviewer | What to include |
|---|---|---|
| End-to-end capstone | Can you frame and communicate an analysis? | Question, source, cleaning, result, limitation, recommendation |
| SQL investigation | Can you explain and validate table relationships? | Keys, join checks, runnable query, expected output |
| Spreadsheet report | Can someone maintain your calculations? | Raw tab, formulas, checks, definitions, refresh notes |
| Visualization | Can an audience interpret the result? | Clear units, filters, captions, accessible design |
The current Google course sequence includes Python as well as SQL, spreadsheets, and Tableau. Include a Python notebook only if you can explain its transformations and reproduce its result. An older R capstone is still legitimate work when identified as legacy-course or independent R practice; do not rename its language to match a newer syllabus.
What belongs in a project README?
Open with the decision and result, then make the evidence easy to inspect. A useful structure is: business question; dataset and license; scope and dates; methods; key finding; recommendation; limitations; reproduction steps; file guide. The opening should tell a reviewer why the project exists before they have to download a notebook.
For a synthetic delivery-analysis project, an opening could read: “This practice project compares on-time delivery across two service zones. Zone B has more late deliveries per 100 completed orders in the sample. The report identifies days to investigate, but cannot attribute delays to drivers because route complexity and dispatch times are unavailable.” Those are example words, not claims about your work.
Below the opening, list the actual files and what each does. “cleaning.sql removes confirmed duplicate event IDs and records exceptions” is more useful than “SQL project.” Include the commands, spreadsheet steps, or notebook order needed to regenerate the published totals. If a paid data source cannot be shared, state that limitation and provide an authorized synthetic sample instead.
How do you make a guided case study your own?
Preserve attribution to the course or original assignment. Add a defensible question, a sensitivity check, a clearer validation method, or an audience-specific output. Copying a finished public notebook and changing colors does not establish your own analytical contribution. Explain which decisions you made and which steps came from the learning materials.
For example, if a guided analysis compares total rides by membership type, you might investigate whether the conclusion changes when totals are normalized by the number of observed days. Or compare results before and after excluding a documented incomplete month. The extension must be supported by the dataset; do not add an impressive-sounding question that requires fields you do not have.
Maintain a short decision log. Record why you chose a left join, how you treated missing amounts, and what happened when you changed the date window. These notes become useful interview evidence because they explain reasoning that a final screenshot cannot show.
Where should you publish the work?
Choose a format your intended reader can open. A repository is useful for code and documentation; a simple project page is useful for a narrative; an interactive dashboard is useful when filters materially help interpretation. You do not need to build a custom website before presenting a well-organized case study. Test every public link in a signed-out session.
For a notebook, include a readable rendered result as well as the source when possible. For a spreadsheet, provide viewing access without accidentally granting editing rights. For a dashboard, include a static caption and methodology link so the result still makes sense if the interactive view loads slowly.
Tableau's publishing guidance explains that workbooks saved to Tableau Public are public. Inspect the source data and permitted sharing before uploading. Removing a name column may not fully anonymize records if other fields identify a person. Use public, licensed, or synthetic data that you are allowed to share.
What should you check before sharing?
- Metric consistency: the README, chart, and query should use the same denominator, time window, and units.
- Reproducibility: a fresh run should produce the published totals without relying on hidden notebook state.
- Source clarity: distinguish collected data, downloaded data, and synthetic examples.
- Limitations: identify missing fields, incomplete periods, selection bias, and unsupported causal interpretations.
- Accessibility: use descriptive links, readable labels, and text that does not depend only on color.
Perform a cross-file spot check. If a chart says 180 completed orders, locate the query or formula that produces 180. If a cleaning log removes ten duplicates, verify that the row count decreases by ten. If the final recommendation cites revenue, make sure the underlying field measures revenue rather than order count.
How can you test whether the portfolio is understandable?
Give someone the opening page without explaining it first. Ask them to identify the question, main result, data source, and biggest limitation. If they cannot find one of those, revise the navigation or wording. This is a usability check for the project, not a prediction of hiring outcomes.
Next, rehearse three levels of explanation. In 30 seconds, name the decision and result. In two minutes, explain the data and a difficult choice. In ten minutes, walk through a query or formula and its checks. Be prepared to say which part you would improve with more data instead of defending every initial decision as perfect.
When should you build another project?
Add another project when it closes a specific evidence gap. If target postings ask for SQL joins and your portfolio contains only charts, a query-focused investigation is useful. If your code is strong but every explanation is technical, write a concise stakeholder memo before learning another visualization package. Keep an archive for older experiments so unfinished work does not obscure your strongest examples.
Verified October 14, 2026: capstone/portfolio scope checked against Coursera and public-sharing behavior against Tableau documentation. Portfolio selection and review steps are editorial recommendations; no hiring success rate is claimed.
Practice next
Use the Google Data Analytics Certificate overview to place this topic in the current course sequence. Use the worked capstone plan for an initial project and the resume and LinkedIn guide to describe it accurately.
The Google Data Analytics study guide PDF includes portfolio, capstone, resume, and LinkedIn templates. The 242-page expanded edition includes current Python labs, 120 multiple-choice questions, 69 applied prompts, and companion practice files. It is an independent supplement to the official coursework. For questions as you practice, explore the separate SimpuTech Google Data Analytics study coach.