A practical DP-800
study plan.
Use this plan to turn reading into decisions you can explain. It is designed for learners who already have basic SQL experience. Spread the sessions across your available weeks; a beginner may need extra time for the foundations.
Build a schedule for my available time →Before you start: set a baseline
Start with a small diagnostic and record your confidence before checking each answer. A wrong answer you felt certain about deserves attention before a topic you already know is weak. The free DP-800 sample follows this method and includes a Microsoft source for every explanation.
Keep one working database for your exercises, one notebook for decisions, and one revision tracker. Aim to leave each session with a query, a tested change, or a diagram you can explain. Reading another page is useful only when it resolves a specific gap.
Microsoft’s published domain ranges are 35–40% design and development, 35–40% security, optimization and deployment, and 25–30% AI capabilities. Use the domains to distribute your study, then adjust for your own gaps.
Your 14-session plan
For a 45-minute session, try 10 minutes of targeted reading, 25 minutes of hands-on work, and 10 minutes explaining the decision and writing an error note. This is our suggested study routine; change the timing to fit your experience.
| Session | Focus | What to produce |
|---|---|---|
| 1 | Baseline & scope | Take the free diagnostic. Mark every wrong answer you felt sure about. Read the official skills outline before choosing a weak area. |
| 2 | Schema decisions | Build a small orders schema. Write down the reason for each data type, key, constraint, and index. |
| 3 | Query logic | Write a query with a window function and a CTE. Create a tiny test dataset and predict the result before running it. |
| 4 | JSON & validation | Turn an array of order lines into relational rows. Test missing properties, wrong types, and an empty array. |
| 5 | Programmability | Put a business operation in a stored procedure. Test its permissions and error behavior instead of checking only the happy path. |
| 6 | Transactions | Create an error inside a transaction. Explain which changes should survive and how you know the session is safe afterward. |
| 7 | Data protection | Compare row filtering, masking, and encryption for a tenant application. Identify the threat each control actually addresses. |
| 8 | Performance evidence | Inspect a slow query. Record what the plan shows, your proposed change, and the measurement that would prove improvement. |
| 9 | Deployment discipline | Review a SQL database project change. Identify drift, destructive changes, secrets, and rollback considerations before release. |
| 10 | Database integration | Trace a database API request from identity to query result. Check how errors and row permissions reach the caller. |
| 11 | Embedding lifecycle | Sketch an embedding refresh process. Decide what happens when the source text changes or generation fails. |
| 12 | Retrieval choices | Compare exact and approximate vector retrieval. Explain when a keyword signal should also influence the result. |
| 13 | RAG & boundaries | Trace a retrieved passage into a model response. Identify what comes from data, what comes from the model, and which access controls apply. |
| 14 | Repair & reassess | Revisit your error log before another practice set. Explain the deciding constraint and why each distractor fails without reading the answer. |
Fourteen sessions do not guarantee readiness or complete coverage. Use the official skills outline as a checklist, add missing topics, and spend extra sessions where your evidence is weakest.
Turn mistakes into a short revision queue
- Write the deciding constraint. Describe the requirement that eliminates the tempting wrong option.
- Keep the official source beside it. Record the documented behavior instead of memorizing a letter.
- Test the concept in another setting. Change the dataset, permission, or requirement and predict the outcome.
- Recheck after a gap. Return later and explain the answer without looking at your note. Familiarity with a repeated question is different from understanding a new scenario.
Questions learners ask
Can I use this plan without buying a course?
Yes. The guide, planner, and tracker are free. Link your tasks to Microsoft Learn and use the free question sample to diagnose gaps. The separate 40-question practice product is optional.
How long should I study for DP-800?
That depends on your SQL experience, available time, and diagnostic gaps. Start with a realistic weekly schedule, review your results, and add sessions where you cannot yet explain or implement the concept. A fixed number of days is not a pass prediction.
Does this include real exam questions?
No. Prep With Proof publishes original practice scenarios. Microsoft’s confidential exam questions are not part of this guide or our question sets.
What should I do if the blueprint changes?
Compare Microsoft’s change log with your notes and schedule. Add the changed objectives to your practice queue, and verify product-specific behavior in the current official documentation.