Power BI for Beginners: Building a Dashboard That Belongs in a Portfolio

Power BI appears in 49.2% of UK analyst adverts, ahead of Tableau. Here is how to build a first dashboard that survives an interview.

Power BI appears in 49.2% of UK data analyst job adverts, ahead of Tableau at 39.5%, and the surrounding Microsoft ecosystem widens that lead further. For UK roles specifically, it is the BI tool to learn first.

Most beginner dashboards fail for the same reason: they are a collection of charts rather than an answer to a question. Here is how to build one that survives an interview.

Step 1: Start with the question, not the data

Before you download anything, write one sentence: the question your dashboard answers.

"Which of our five regions is underperforming on repeat purchase rate, and is it a customer acquisition problem or a retention problem?" is a question. "Sales dashboard" is not.

This single discipline separates portfolio pieces from practice files, and it is the first thing an interviewer will probe.

Step 2: Choose data with real mess in it

Good UK sources: data.gov.uk, NHS Digital, ONS, TfL open data, DVSA, or Kaggle for commercial-style datasets.

Choose something with a few thousand rows minimum, at least two related tables, and genuine imperfections. Perfectly clean data produces a dashboard that demonstrates nothing, because it skips the part of the job you are trying to prove you can do.

Pick a sector you want to work in. A housing or NHS dataset is a talking point when applying to a housing association or trust.

Step 3: Clean in Power Query, and document it

Power Query is where the actual analyst work happens, and it is the step beginners skip.

Remove duplicates, set data types explicitly, handle blanks deliberately, standardise inconsistent categories ("London", "london", "Greater London"), split and merge columns as needed, and unpivot data that arrived in a wide layout.

Crucially: keep a note of every decision and why. "Excluded 340 rows with null transaction dates, representing 0.8% of the dataset, after confirming they were test records" is exactly the kind of sentence that makes an interviewer sit up.

Step 4: Model the data properly

This is the step that most distinguishes a competent dashboard from a beginner one, and the step where DAX problems actually originate.

Build a star schema: one central fact table containing the events or transactions, surrounded by dimension tables for things like date, product, customer and region. Set one-to-many relationships from dimensions to the fact table.

Always add a dedicated date table. Time intelligence functions in DAX depend on it, and almost every dashboard needs comparisons over time.

Do not merge everything into one flat table because it feels simpler. It works at 500 rows and falls apart at scale, and it is visible to anyone who opens your file.

Step 5: Write the DAX that matters

You need far less DAX than the internet suggests. For a portfolio dashboard:

Basic aggregations: SUM, AVERAGE, COUNTROWS, DISTINCTCOUNT
CALCULATE, which is the function everything else is built on
Time intelligence: SAMEPERIODLASTYEAR, DATEADD, TOTALYTD
DIVIDE for safe division
One or two measures using variables (VAR/RETURN) to show you can write readable code

Understanding filter context matters more than knowing many functions. If you can explain why a measure returns a different value inside a visual than it does in a card, you understand DAX better than most beginners.

Step 6: Design for one screen and one message

Restraint reads as competence. Cramming reads as insecurity.

A strong single-page layout: two to four KPI cards at the top for headline numbers, one trend chart, one breakdown by category, one detail table, and a small number of slicers. Consistent colours with one accent reserved for the thing you want noticed. Titles that state findings ("Repeat purchase rate down 12% in the North East") rather than describing charts ("Repeat purchase by region").

Step 7: Publish it, and write the two paragraphs underneath

Publish to the Power BI service and generate a shareable link, or record a short walkthrough if the data cannot be public.

Then write the part that turns a file into a portfolio piece:

Paragraph one: the question, the data source, and the key cleaning decisions. Paragraph two: what you found, and what you would recommend someone do about it.

Without those two paragraphs, an employer sees charts. With them, they see an analyst.

Should you take PL-300?

Microsoft's PL-300 (Power BI Data Analyst Associate) is the certification most directly matched to what UK adverts ask for, covering data preparation, modelling, visualisation and asset management. Azure certifications as a category appear in 34.4% of UK data analyst adverts.

It is a genuine signal, particularly for career changers with no relevant employment history, and it is renewable free online each year. It is not a substitute for a portfolio. The combination of the two is what works: the certification gets you past screening, the dashboard gets you through the interview.

Frequently asked questions

Do I need Power BI Pro to build a portfolio dashboard? Power BI Desktop is free and sufficient for building. Publishing to the service and sharing may require a licence depending on your setup, so a recorded walkthrough is a reasonable fallback.

How many dashboards should I have? One excellent one beats three average ones. If you build more, make each answer a different type of question rather than repeating the same pattern on different data.

Power BI or Tableau? For the UK, Power BI: 49.2% of adverts versus 39.5%, and the Microsoft ecosystem reinforces it. The concepts transfer if you later need Tableau.

Uptrail's Power BI module works through this exact process on real business scenarios, and PL-300 is the external certification built into our AI Data Analyst Career Programme, because it is the credential UK job adverts most consistently recognise.

Sources: ITJobsWatch, Data Analyst skills co-occurrence data, 6 months to 1 September 2026; Microsoft Learn, PL-300 certification skills outline.