How to Build a Data Analytics Portfolio Recruiters Notice

What recruiters actually assess in a portfolio

Data recruiters read a portfolio to answer one simple question: can this person turn messy data into a decision. They look for the trail of your thinking, from framing the question, choosing the data, and cleaning it, to the final conclusion. A row of certificates only shows you attended a course. A finished project shows you can carry real work over the line. That is why a strong portfolio rests on project quality and clear documentation. Recruiters want to see tidy SQL queries, a dashboard that answers one question decisively, and a summary explaining why the finding matters to the business. The ability to narrate your process is as valuable as the result itself. Local context adds even more value. Analysing Jakarta transport data, food prices in Indonesian markets, or national online-shopping trends makes a project feel relevant to companies here. Government open data through the Satu Data Indonesia portal offers fresh, meaningful material to work with.

7 Steps to Assemble a Data Analytics Portfolio From Scratch

These seven steps guide you from a blank page to a tidy portfolio ready to link on your CV. Work through them in order, and prioritise depth in each project over simply adding more.

1. Step 1

   Decide on three varied project themes

   Start by designing your portfolio composition instead of downloading random datasets. Choose three themes that show different skills: one exploration and visualisation project, one that foregrounds multi-table SQL queries, and one heavy data-cleaning project ending in a dashboard. This range gives recruiters a full picture of your reach. Tie each theme to a personal interest, such as football, music, transport, or household finance, so you stay motivated to finish. This early blueprint keeps the portfolio focused and stops it from becoming an aimless pile of files.

   Tips

   • Reserve one theme close to the industry you are targeting, such as retail or banking
   • Limit yourself to three themes first so each project gets finished properly
2. Step 2

   Find clean, meaningful public datasets

   Dataset quality decides half of a project's success. Pull data from trusted sources such as Kaggle, the Satu Data Indonesia portal, Statistics Indonesia, or open city-data repositories. Choose a dataset large enough to reveal patterns yet small enough that you understand its context. Check the usage licence and confirm the data may be republished in a portfolio. Avoid overly popular datasets worked by thousands of people the same way, such as the Titanic passenger data, because the results struggle to stand out. A rarely-touched local dataset makes your analysis feel fresh to Indonesian recruiters.

   Tips

   • Read the column descriptions carefully before committing to a dataset
   • Save the source link so it is easy to cite in your documentation later
   Check every dataset's licence. Some data may only be used for personal purposes and must not be republished publicly.
3. Step 3

   Frame one sharp business question

   A project without a question turns into a pile of aimless charts. Before touching the data, write one concrete question to answer, such as which region's sales dropped over the last three months, or what time a transport service is busiest. A good question is specific, measurable, and relevant to a decision maker. From that main question, derive two or three supporting sub-questions. This frame becomes the backbone of your analysis and the storyline you will tell in the dashboard and summary. Recruiters value candidates who sharpen the problem before computing anything.

   Tips

   • Picture the decision maker who will use your answer
   • Test your question by asking, if answered, what decision would change
4. Step 4

   Run the full workflow from cleaning to findings

   This is where your core skill shows. Import the data, then clean the messy parts: missing values, inconsistent date formats, duplicates, and mixed units. Record every cleaning decision, because recruiters often ask about them. Once the data is tidy, run the analysis with SQL or Excel to group, join, and compute the summaries that answer your question. Look for patterns, compare groups, and trace interesting anomalies. This stage displays both your technical ability and your care. Honest analysis states the data's limitations, so the conclusion can be trusted.

   Tips

   • Keep the raw and cleaned data separate so the process stays traceable
   • Comment your SQL queries so your reasoning is easy to reread
   Do not delete outliers without a reason. Investigate the cause first, because outliers often hold the most interesting findings.
5. Step 5

   Build a dashboard that answers the question cleanly

   Turn your findings into a dashboard readable in seconds. Choose the right chart type: bars to compare categories, lines for time trends, and maps for regional data. Place the most important number at the top, then arrange supporting charts to follow the storyline of your question. Give titles that explain the finding instead of just naming the chart. Publish the dashboard through Tableau Public or Power BI so recruiters can open it straight from a link. A clean, focused dashboard shows your ability to convey results to non-technical people, a skill highly valued at work.

   Tips

   • Limit your colour palette so the dashboard stays calm and the message stays clear
   • Ask a layperson to read your dashboard, if they understand it in a minute, the design works
6. Step 6

   Document each project as a case study

   Documentation turns files into a story that can be assessed. For each project, write a fixed-format summary: background and question, data source, cleaning and analysis steps, key findings, and the recommendations that emerged. Include dashboard screenshots and a link to the code. Save this write-up as a README on GitHub so anyone opening the repository immediately grasps the project's core. Clear, orderly writing reflects how you communicate at work. A good case study answers the questions a recruiter has not yet had a chance to ask.

   Tips

   • Use one shared summary template for every project so the portfolio looks consistent
   • Write findings in business terms, for example eastern-region sales fell 18 percent
7. Step 7

   Assemble everything on GitHub and link it to your CV

   Collect all projects in one tidy GitHub profile, complete with a profile page that introduces you and links each repository. Give repositories descriptive names and arrange files logically: data, notebooks, queries, and documentation kept neatly apart. Link the Tableau Public dashboard in each project's README. Once tidy, place the portfolio link at the top of your CV and LinkedIn profile so it is easy to find. Practise explaining each project out loud too, because data interviews often test how you reason behind the numbers. An accessible portfolio multiplies your chances of getting noticed.

   Tips

   • Pin your three best repositories on your GitHub profile so they show first
   • Prepare a one-minute story for each flagship project ahead of interviews

Four Public Dataset Sources for Portfolio Projects

A Strong Portfolio Versus a Weak One

The biggest difference lies in clarity of story. A project that can explain its question, process, and impact will always beat a pile of context-free files.

Portfolio Readiness Checklist Before Applying

Building a Portfolio Alone or With Guidance

Frequently Asked Questions