70-768 Developing SQL Data Models: Candidate Guide and Current Status
Exam 70-768 was titled “Developing SQL Data Models” and was associated with Microsoft’s SQL Server data and MCSA SQL 2016 BI Development pathways. It was intended for candidates building or extending SQL-based data-modeling expertise within those historical certification routes. The immediate decision for a reader today is not whether to book it, but whether the exam is still available, what its title and pathway placement imply for study, and which current Microsoft option should replace it.
What did 70-768 validate?
The official evidence identifies 70-768 as “Developing SQL Data Models.” Its placement in Microsoft certification-path material connects it with SQL Server data certification and the MCSA SQL 2016 BI Development pathway. That establishes the exam’s subject direction, but the supplied sources do not provide a current skills outline, objective list, scoring method, or detailed examination blueprint.
A useful interpretation is that the exam focused on the design and development of SQL data models rather than general SQL administration or basic query writing. That interpretation should guide historical study, but it should not be treated as a substitute for an official objective document. The available evidence does not support assigning particular percentages to domains or claiming that a specific feature appeared on the exam.
The exam’s pathway placement also matters. A candidate could encounter 70-768 as part of a broader Microsoft data-certification plan rather than as an isolated database test. The supplied Microsoft Q&A material lists it among elective exams associated with the former MCSE: Data Management and Analytics route, while the certification-path PDFs associate it with SQL Server data pathways. Those historical relationships should be checked against the candidate’s transcript and original certification goal, not assumed to describe a current credential.
Is 70-768 still available?
No. Microsoft’s official material states that 70-768 was among the exams scheduled for retirement, and Microsoft’s current retirement guidance says a retired exam cannot be taken after its retirement date. The Microsoft Q&A page also records the exam as retired on January 31, 2021. A new candidate should therefore not plan a booking, purchase preparation material advertised as a live exam, or expect to earn a current certification through 70-768.
The historical retirement announcement contains more than one retirement reference because Microsoft updated its communication during the transition from legacy certifications to role-based certifications. For this exam, the supplied Q&A evidence explicitly reports the retirement date as January 31, 2021, and the retirement-policy page explains the practical consequence: the exam and associated credential are no longer available after retirement. Use Microsoft’s retirement page as the controlling policy source if transcript or legacy-certification questions arise.
Microsoft explains that certifications already earned remain on a Microsoft Learn transcript after the associated exams retire. That is different from earning the certification now. A candidate who passed the required exams before the applicable retirement point may retain the credential record, but a person who has not completed the requirements cannot use a retired exam as a present-day route.
What does retirement mean for an existing credential?
Retirement does not erase a credential that was already earned. Microsoft states that an existing MCSA, MCSD, or MCSE certification remains on the candidate’s transcript after the exams retire. The retirement policy likewise says an already-earned certification or Applied Skills credential stays on the Microsoft Learn profile. This is a transcript-status issue, not permission to sit 70-768 again.
If you are checking an old certification plan, first sign in to the Microsoft certification profile associated with your historical exams and review the transcript. Confirm which exams were passed, which certification requirements were complete, and whether the credential was actually awarded. Do not infer completion from a training course, an exam voucher, or an unofficial pathway diagram.
Can a passed exam still matter to a historical pathway?
Microsoft’s retirement announcement states that if qualifying exams were passed before retirement, they could continue to count toward competency attainment for a limited post-retirement period. Because that transition treatment was tied to historical program rules, candidates should verify their individual transcript and the relevant Microsoft support guidance rather than applying the rule to a new attempt. A failed or unattempted 70-768 cannot be converted into a current exam booking.
Who was the exam for?
The strongest evidence-based audience is a SQL Server or business-intelligence practitioner working toward a historical Microsoft data certification. The title points to data-model development, while Microsoft pathway documents connect the exam with SQL Server data certification and MCSA SQL 2016 BI Development. It is not appropriate to describe it as a current role-based exam for a specific modern job title because the supplied sources do not make that claim.
Historically, the exam would have been relevant to candidates who needed a data-modeling elective or who were completing a SQL Server business-intelligence development sequence. The Microsoft Q&A response lists 70-768 alongside other former electives for MCSE: Data Management and Analytics, including database development, data warehousing, reporting, and Azure Cosmos DB topics. That list shows the breadth of the old pathway; it does not prove that every listed technology belonged to 70-768 itself.
A sensible audience assessment is therefore based on the candidate’s original objective. If your goal was a legacy SQL 2016 BI Development credential, 70-768 belonged to that plan. If your goal is a currently recognized data-platform certification, the retirement status changes the decision: use the old title to understand your prior learning target, then investigate an available Microsoft certification rather than searching for a replacement exam by assumption.
What official exam details are not available here?
The supplied official research does not provide an authoritative 70-768 skills-measured page, domain percentages, question count, exam duration, passing score, language list, price, prerequisites, or delivery method. Those details should not be copied from third-party pages as though they were verified. The safest current statement is that the exam is retired and that the available evidence supports only its title, historical pathway placement, and retirement consequences.
There are no supported blueprint weights to reproduce. In particular, do not attach percentages from another SQL, BI, Azure, or role-based exam to 70-768. A percentage is meaningful only when paired with the exact domain named in its official blueprint, and no such 70-768 blueprint appears in the supplied sources.
The same caution applies to scheduling. Microsoft’s retirement policy says candidates should take an exam before its retirement date if they are preparing for one, but that general guidance does not create a present-day appointment option for 70-768. Current registration, delivery, and regional availability should be checked only on Microsoft’s live certification and exam pages; the supplied research does not evidence any current delivery channel for this exam.
How should you treat old objective lists?
Use an archived objective list, if you already possess one, as a historical study reference rather than a live exam contract. Compare its concepts with the work you want to perform today, and label notes as historical. Do not assume that a blog’s list of question topics, claimed exam weights, or alleged test format came from Microsoft unless the official source directly supports it.
How should you prepare if your goal is historical understanding?
Build preparation around working data-model decisions, not memorized answer strings. Start with relational modeling and dimensional concepts, then implement a small SQL-based model, test its behavior with representative queries, and document why you chose each structure. This approach is a practical recommendation for learning the subject; it is not a claim about an undisclosed 70-768 blueprint or a route to taking the retired exam.
A productive study environment should let you create tables, keys, relationships, constraints, views, and representative analytical queries. Use a small but coherent scenario such as orders, customers, products, and dates. The scenario is a study exercise, not a reported exam case. The point is to make modeling trade-offs visible: grain, duplicate data, relationship direction, null handling, history, and query usability.
Keep a decision log. For each model change, record the business question, the grain of the fact or transaction, the dimensions involved, the integrity rule being protected, and the query pattern being improved. This turns passive reading into evidence that you can reason about a model. It also exposes gaps more reliably than repeatedly reviewing terminology.
Stage one: establish modeling foundations
Begin by separating three layers of thinking: the business process, the logical data model, and the physical SQL implementation. Define the entities and events before writing DDL. State what one row represents in each important table. If you cannot explain a table’s grain in one sentence, postpone optimization and clarify the design first.
Review primary and foreign keys, candidate keys, relationship cardinality, nullability, constraints, and normalization principles. Then ask when a reporting model may deliberately reshape data for analysis. The objective is not to choose normalization or denormalization automatically; it is to justify the choice against integrity, maintainability, and query requirements.
Practice by taking a deliberately flawed model and identifying update anomalies, duplicated attributes, ambiguous relationships, and missing constraints. Redesign it, implement the change, and run queries that demonstrate the difference. Keep both versions so you can explain the consequence of each decision.
Stage two: implement and inspect a model
Translate the logical design into SQL objects and test the implementation with realistic inserts, updates, and deletes. Confirm that constraints reject invalid relationships and that the model behaves correctly when optional data is absent. Read the result sets carefully; a query that returns rows is not necessarily proving that the model is correct.
Add analytical queries that cross the main relationships. Inspect whether joins multiply rows unexpectedly, whether aggregations use the intended grain, and whether filters produce consistent results. Use execution plans or other available diagnostic tools to investigate inefficient patterns, but do not optimize before verifying correctness.
Create a change exercise: add a new attribute, introduce a historical version of an attribute, or support a new reporting question. Note whether the change requires table redesign, a new relationship, a bridge structure, or only a query adjustment. The exercise develops the design judgment implied by “developing data models” without pretending to reproduce retired exam content.
Stage three: connect models to BI work
If your original target was MCSA SQL 2016 BI Development, study the boundary between model development and reporting use. A model must preserve trustworthy relationships while remaining understandable for analytical queries. Test common dimensions, measures, date filtering, aggregation, and many-to-many cases in a controlled dataset.
Compare a normalized transactional design with a reporting-oriented design built from the same business process. Explain which design better serves writes, which supports recurring analysis, and where extra transformation or governance is required. This comparison is more valuable than labeling one design universally superior.
Finish each lab by writing a short review: what the model guarantees, what it does not guarantee, which assumptions it contains, and what a future developer must know before extending it. Clear reasoning is the transferable skill left after the retired exam itself is no longer a scheduling target.
Which study mistakes waste the most time?
The largest mistake is preparing as if 70-768 were still an active exam. Confirm status before buying a course, booking travel, or building a deadline around it. The second is treating an unofficial topic list as a Microsoft blueprint. Because the supplied evidence contains no current objectives or weights, preparation should focus on durable modeling practice and on selecting a live certification path.
Do not confuse adjacent exam titles. Microsoft’s historical material lists 70-767 as implementing a data warehouse using SQL and 70-768 as developing SQL data models. They were related within the BI pathway, but their titles describe different emphases. Studying warehouse implementation alone does not demonstrate model-development competence, and studying model design does not establish that you covered every warehouse topic.
Do not use dumps, leaked questions, or answer memorization as a preparation strategy. They cannot establish that you understand keys, grain, integrity, relationship behavior, or maintainable design, and they are especially unreliable when an exam has retired. Build and test models instead.
Finally, avoid making a certification decision from an old diagram without checking your transcript. Historical pathways changed, and a forum response may describe an earlier requirement. Treat community material as a lead to verify, not as proof that a current credential can still be earned.
A quick quality check for your notes
Every note should answer one of three questions: what problem does this modeling choice solve, what trade-off does it introduce, or how can I verify its behavior in SQL? Remove notes that are only lists of remembered answers. Mark claims about exam logistics separately from technical notes, and discard any logistics claim that cannot be tied to an official Microsoft source.
What is the practical study roadmap?
Use a four-part roadmap: verify the credential goal, build a model, test and explain it, then choose a current certification path. This sequence prevents a common failure—spending weeks studying a retired exam before checking whether the intended credential is still attainable. The roadmap below is a practical recommendation, not an official 70-768 schedule or Microsoft requirement.
First, make a status decision. Read Microsoft’s retirement guidance, inspect your transcript, and write down whether you are documenting a credential already earned, completing historical records, or pursuing a current alternative. If the answer is “current alternative,” stop treating 70-768 as a booking target.
Next, create one end-to-end data-model project. Define a business process and grain, design the logical model, implement the SQL objects, load representative data, and run integrity and analytical tests. Do not expand the project endlessly. A small model that you can defend is better preparation than a large, undocumented schema.
Then perform a review without consulting notes. Explain every key and relationship, predict the result of a join before running it, diagnose an intentionally flawed query, and propose a controlled schema change. Record errors by concept rather than by isolated question. For example, classify a wrong result as a grain, relationship, nullability, or aggregation problem.
Finally, map the skills you still need to a currently available Microsoft learning and certification option. Microsoft’s retirement communications explain that role-based certifications replaced parts of the older program and that SQL Server content could appear in role-based certifications on an as-needed basis for certain job roles. That is a direction for research, not evidence that one specific current exam replaces 70-768.
A readiness test for the technical subject
You are ready to move on from introductory study when you can state a table’s grain, defend its keys, explain its relationships, demonstrate how constraints protect it, and identify how a query could produce duplicate or misleading aggregates. You should also be able to change the model for a new business question and describe the cost of that change. These are practical indicators, not an official passing standard.
A readiness test for the certification decision
You are ready to schedule a different exam only after confirming its current status, official objectives, prerequisites, delivery information, and registration route on Microsoft’s live pages. The supplied sources do not provide those details for a replacement. Do not assume that DP-300, another SQL exam, or a role-based credential is an automatic substitute; select it by the job skill and certification requirement you actually need.
What should you do next?
If you planned to take 70-768, do not search for a test appointment first. Verify whether you already passed it, identify the historical certification it supported, and check Microsoft’s current catalog for an available credential aligned with your work. If your purpose is technical learning, use the title as a starting point for a hands-on SQL data-model project and validate the resulting skills against current documentation.
For a transcript or legacy-path question, use Microsoft’s certification support and profile resources rather than relying on a third-party listing. For a replacement decision, compare current role-based certification objectives with your responsibilities in data modeling, database development, analytics, or administration. The Microsoft Q&A response about replacement exams records that no replacement was provided in that discussion; it should not be read as a permanent catalog statement, so confirm the live catalog before acting.
The central decision is straightforward: 70-768 can still be useful as a historical label for SQL data-modeling study, but it is not a current exam target. Separate the value of the technical subject from the availability of the credential, and spend your preparation time on skills and certification requirements that you can verify today.
Conclusion
Exam 70-768 belongs to Microsoft’s retired SQL Server and BI certification history. The official sources support its title, pathway associations, and retirement consequences, but they do not support current blueprint weights or delivery details. Candidates should verify their transcript, avoid unsupported replacement claims, practice data modeling through implementation and testing, and choose a current Microsoft credential only after checking its live requirements and availability.
