A data analyst's portfolio shows how a question became a decision. Build two or three case studies that each state the business question, the data and methods you used, the key finding in one chart and what the business did with it. Reviewers care more about clear reasoning than complex models.
For: Data analysts who want to show how their analysis led to business decisions
Open each case with the question a team needed answered, such as why retention dropped or which customers to target, and the number that raised it. This tells the reviewer why the analysis mattered before they see any charts. Keep it to one or two sentences. If you found the question yourself, say how.
Describe the data sources, their size and the main methods, such as cohort analysis, segmentation, regression or an A/B test readout. Name the tools you used, two to four per case, such as SQL, Python or a BI tool. Mention any cleaning or assumptions that affect the result. Link a notebook or a public dashboard if the data can be shared.
Pick the single chart that carries the finding, give it a title that states the insight, and caption it with the takeaway. Then show what the business did with it and what changed in numbers afterwards. An analysis that led to an action is more convincing than a dashboard nobody used. State your role and who you worked with, such as a PM or a marketing team.
BeforeAnalyzed customer data using SQL and Python.
AfterRan a cohort analysis on 18 months of order data in SQL, finding that first-order delivery delays doubled 90-day churn.
The method, the scope and the finding show analytical skill; tools alone do not.
BeforeCreated dashboards for the marketing team.
AfterBuilt a weekly channel dashboard for a marketing team of 6, replacing 4 manual reports and saving about 5 hours a week.
Naming the users and the time saved shows the dashboard was actually adopted.
BeforeProvided insights that helped improve retention.
AfterIdentified 3 at-risk segments and recommended a win-back offer that lifted 60-day retention in those segments by 12%.
Connecting the insight to an action and a measured change shows the analysis mattered to the business.
Yes, especially when work data is confidential or you are changing careers. Choose a dataset with a realistic business question, and structure the case like work: question, data, method, finding and a recommendation. Link the notebook so the reviewer can check your code.
Write for a hiring manager first and a technical reviewer second. Keep the main slides to the question, finding and decision, and put SQL snippets, model details or validation steps in an appendix or linked notebook.
One key chart per case is often enough, with one or two supporting charts at most. Each chart should have an insight title and a caption. Too many charts make the reader search for the point.
The button opens a portfolio filled with this example in the editor. You need to sign in; change the text, images and layout of every slide freely and download it as PDF or PPTX. Drop your own work onto the image spots.
Start with this example ↗