Lesson 1: Exam Orientation and Ways to Represent Data

Before you start: the facts about the exam

This course is made by LevelUpWith. LevelUpWith is independent of Microsoft. It is not endorsed by Microsoft, and nobody can promise you a pass. These are the facts Microsoft publishes about Exam DP-900:

1. Duration: Microsoft says you have 45 minutes to complete the exam. If the exam is not offered in your preferred language, you can ask for an extra 30 minutes.
2. Pass mark: you need a score of 700 or greater to pass.
3. Study guide: the official study guide belongs to Microsoft. It is free on Microsoft Learn (search for "Study guide for Exam DP-900"). This course does not reproduce it, but it follows its "skills measured as of July 21, 2026".
4. Practice assessment: the study guide links to a free Practice Assessment on Microsoft Learn. Microsoft also offers an exam sandbox, where you can try the exam's question types before the real thing.
5. Retakes: if you fail, you can retake the exam 24 hours after your first attempt. Later retakes have longer waits.

DO now: open Microsoft Learn, find the DP-900 study guide and bookmark it. Bookmark the free Practice Assessment as well. You will use both throughout this course.

How the exam is weighted

The study guide splits the exam into four skill areas:

• Describe core data concepts: 25–30%
• Identify considerations for relational data on Azure: 20–25%
• Describe considerations for working with non-relational data on Azure: 15–20%
• Describe an analytics workload on Azure: 25–30%

This level starts the first area with the skill "Describe ways to represent data". It covers the features of structured, semi-structured and unstructured data.

Key vocabulary

• Data: facts such as numbers, descriptions and observations that record information.
• Entity: something that matters to the business, such as a customer, a product or a sales order.
• Attribute: a characteristic of an entity. A customer's attributes might be a name, an address and a phone number.
• Schema: the agreed layout of the data, meaning which fields exist and what each one holds.

Structured data: a fixed schema

Structured data follows a fixed schema, so every record has the same fields.

Business example: a sales table.

• Each row is one instance of the entity (one sale).
• Each column is one attribute (date, quantity, and so on).
• "Fixed schema" means the columns are set in advance. You cannot give one order an extra "GiftMessage" column without changing the whole table.
• Structured data is most often tabular and is usually stored in a database. In a relational model, tables refer to each other through key values. For example, ProductID points to a Product table. Relational data is covered in a later level.

Analogy: a printed form with fixed boxes. Every customer fills in the same boxes.

Semi-structured data: key-value fields and nesting

Semi-structured data has some structure but allows variation between records. A common format is JSON (JavaScript Object Notation).

Business example: a JSON customer profile.

{
  "firstName": "Priya",
  "lastName": "Shah",
  "address": {
    "city": "Leeds",
    "postalCode": "LS1 4AP"
  },
  "contact": [
    { "type": "email", "address": "priya@contoso.com" },
    { "type": "mobile", "number": "07700 900123" }
  ]
}

Three terms to know:

1. Key-value fields. Each piece of data carries its own label. In "firstName": "Priya", firstName is the key (the field name) and Priya is the value. Because each record labels its own data, the structure travels with it.
2. Nested entities. One entity can sit inside another. Here address is an object inside the customer, and contact is a list of contact entries.
3. Variation. A second customer might have three email addresses, no mobile number, or an extra unit field in the address, and the data is still valid. A fixed-schema table could not hold this without changing its design.

Analogy: a labelled folder. Every item inside has a name tag, but folders do not all hold the same items.

Unstructured data: no defined fields

Unstructured data has no specific structure of fields. Examples include documents, images, audio, video and binary files.

Business example: a scanned invoice (a PDF or JPEG image).

• A person can read the supplier name and the total.
• To a computer it is only pixels. No labelled "Total" field exists until software (such as AI document processing) pulls one out.
• Other examples: recorded support calls, product photos, contracts written in Word, CCTV footage.

Microsoft's training also points out that organisations increasingly work with vector data (also called embeddings). These are numerical representations of content such as text or images, and AI applications use them to find content that is similar in meaning.

How to classify any data example

Work through these steps in order:

1. Ask: does every record have exactly the same predefined fields? If yes, the data is structured. Example: the sales table.
2. If not, ask: does each record carry labelled fields (keys, tags) that may vary or nest? If yes, the data is semi-structured. Example: the JSON profile.
3. If there are no labelled fields at all, the data is unstructured. Example: the scanned invoice.
4. Justify your answer by naming the feature you used: fixed schema, key-value labels or nesting, or no defined fields.

DO now: classify these three examples. A spreadsheet of employee IDs and salaries. A product catalogue in JSON where only some items have a "colour" field. A voicemail recording. (Answers: structured, semi-structured, unstructured.)

What the exam asks about this

• Short scenarios such as "A company stores X. What type of data is this?" Look for the clue word: table / rows / columns / fixed fields points to structured; JSON / fields vary / nested points to semi-structured; images / video / PDFs / audio points to unstructured.
• Questions on the features themselves. A fixed schema belongs to structured data. Variation between records belongs to semi-structured data.
• A common trap: JSON is semi-structured, not unstructured. It does have structure; the structure is just flexible.

Key takeaways

• DP-900 gives you 45 minutes, and you need 700 or greater to pass.
• Microsoft's study guide and Practice Assessment are free on Microsoft Learn. LevelUpWith is independent of Microsoft and cannot promise a pass.
• Structured data has a fixed schema: same fields for every record, usually in tables.
• Semi-structured data uses labelled key-value fields, allows variation and can nest entities. JSON is the classic example.
• Unstructured data has no defined fields: documents, images, audio, video and binary files.