Platform Data Architect Certification: 2026 Exam Guide

The Short Answer

The Salesforce Certified Platform Data Architect exam (Plat-Arch-201) has 60 scored multiple-choice/multiple-select questions plus up to 5 unscored, a 105-minute time limit, a 58% passing score, a US$400 registration fee (US$200 to retake) and no prerequisite. Its questions align to the Spring '23 release (checked in October 2026 against the official Platform Data Architect exam guide and Trailhead Academy).

It's one of the four certifications that together grant Application Architect, which is in turn one of the two prerequisites for Certified Technical Architect. The exam weight sits in three places: Data Modeling/Database Design (25%), Salesforce Data Management (25%) and Large Data Volume Considerations (20%). Expect long scenarios with record counts, ownership patterns and load windows, and answers that turn on how the platform behaves at scale.

Exam Details at a Glance

Sources: the exam guide for format, score, fees, prerequisite, release and maintenance, and Trailhead Academy for the exam code, level and languages, both checked in October 2026.

58% of 60 is 35 correct answers, so you can miss 25. That's the lowest passing score of any exam we've covered on this blog, but don't read it as easy: the questions describe a customer's data problem and offer four plausible designs, and the wrong ones are usually designs that work at 10,000 records and fail at 10 million. 105 minutes for up to 65 questions is about 1.6 minutes each, and the scenarios are long, so practice reading for the numbers first. At US$400 this is a more expensive exam than the US$200 credentials; for fees and retake rules across every Salesforce exam, see Salesforce certification cost and retake fees.

Who the Exam Is For

The exam guide describes a candidate with 2 to 3 years of Salesforce experience and 5+ years supporting or implementing data-centric solutions, in roles such as Advanced Administrator, Data Architect, Technical/Solution Architect or Advanced Platform Developer. They should know custom fields and relationships, the standard object structure for Agentforce Sales and Service, big objects, how license types map to objects, large data volume (LDV) indexing and migrations, and tools such as Data Loader and ETL platforms.

The guide is just as useful for what it says you don't need: non-Salesforce database concepts, configuring integration tools, hands-on experience with master data management (MDM) tools, Lightning development, or a programming language. This is a design exam. You'll be asked which approach to recommend, not to write the code.

There's no required certification, but the guide recommends Platform App Builder, Platform Developer and Platform Developer II first. If you're coming from the declarative side, our Platform App Builder certification guide covers the relationship and object fundamentals this exam assumes.

Where It Sits on the Architect Path

Salesforce doesn't let you sit an "Application Architect" exam. The Application Architect credential page says it's granted automatically once you've passed four prerequisites:

1. Platform App Builder
2. Platform Data Architect
3. Platform Developer
4. Platform Sharing and Visibility Architect

Application Architect is one of the two prerequisites, alongside System Architect, listed on the Certified Technical Architect page. The Platform Data Architect credential page lists Application Architect, System Architect and Technical Architect as its related certifications.

Taking Data Architect after Sharing and Visibility Architect is a sensible order: both exams lean on the same sharing internals (implicit sharing, sharing recalculation, role hierarchy), and Data Architect then asks what happens to them under volume.

Spring '23 Alignment and Data 360

The guide says the questions "align to the Spring '23 release", and its outline never names Data 360. That matters because Salesforce's main answer to "unify customer data from many systems" today is Data 360. Salesforce Help notes that "as of October 14, 2025, Data Cloud has been rebranded to Data 360", with the functionality unchanged, and describes it as a platform to ingest, harmonize and unify data, including identity resolution rulesets and zero copy data federation.

So how should you treat it? The MDM (5%) and single-view-of-the-customer objectives describe concepts Data 360 implements: harmonizing sources, survivorship rules, a unified profile. The Data 360 architecture guide is good background for those ideas. But the exam was written to a Spring '23 platform and is about core platform data architecture: objects, sharing, LDV, archiving and migration. Learn the general MDM vocabulary (golden record, survivorship, canonical model, hierarchy management) and the platform tools first, and treat Data 360 as one option you can recognize in a scenario, not as the default answer to every question. If Data 360 itself is your goal, that's a separate exam: the Data 360 Consultant certification.

The Six Exam Sections and What They Test

All weights and objectives come from the official outline.

Data Modeling/Database Design: 25%

Designing objects, fields and relationships for the Customer 360 Platform; a scalable model that obeys the sharing model; business and technical metadata (business dictionary, lineage, taxonomy, data classification); big objects versus standard and custom objects; and avoiding data skew.

Watch for: the 10,000-record threshold. Ownership data skew is one user owning more than 10,000 records of an object; moving that user in the role hierarchy forces a long sharing recalculation. If you must, keep the owner out of a role (or in a separate role at the top) and out of public groups used by sharing rules. Parent-child skew is 10,000 or more children under one parent, the classic "park every orphan contact under one dummy account" mistake. For big objects, remember they don't support transactions, don't support encryption (archived encrypted data lands as clear text), use custom components instead of standard pages, and need a retry-the-whole-batch write pattern.

Master Data Management: 5%

MDM implementation styles, harmonizing and consolidating sources, survivorship rules, thresholds and weights, reference data for enrichment, canonical modeling, hierarchy management, and establishing a golden record in a single org.

Watch for: "which attribute wins." Survivorship is a per-attribute decision (most recent, most trusted source, most complete), not a per-record one. At 5%, this section is about three questions, so learn the vocabulary and move on.

Salesforce Data Management: 25%

Choosing license types to fit standard and custom object needs, persisting data consistently, a single view of the customer across systems of interaction, and consolidating or using data from multiple Salesforce orgs.

Watch for: the multi-org scenario. Copying records between orgs is rarely the right first answer. The Salesforce Connect cross-org adapter reads data stored in another Salesforce org through external objects, without moving it. For consistency, duplicate management pairs matching rules (how a duplicate is identified) with duplicate rules (what happens when one is found: block the save or just warn).

Data Governance: 10%

A GDPR-compliant data model (identifying, classifying and protecting personal and sensitive data) and designing an enterprise data governance program.

Watch for: what data classification actually stores. The data classification metadata fields are Data Owner, Field Usage, Data Sensitivity Level (Public, Internal, Confidential, Restricted, MissionCritical) and Compliance Categorization (GDPR, CCPA, PII and others). They label a field; they don't encrypt or hide it. Protection comes from field-level security, encryption and the access model, so a "classify the field" answer alone rarely satisfies a "protect the data" requirement.

Large Data Volume Considerations: 20%

A data model that scales with LDV, an archiving and purging plan, and when to use virtualized data.

Watch for: skinny tables. Skinny tables speed up read-only operations (reports, list views, SOQL) on objects with millions of records, but you can't create them yourself: Salesforce Customer Support enables them. They hold up to 200 columns, can't contain fields from other objects, and are copied only to Full sandboxes. For archiving, know the options: big objects for on-platform history, and off-platform stores read back through Salesforce Connect, which is how Legacy Salesforce Archive exposes archived records.

Data Migration: 15%

Data quality at load time, performance techniques for loading LDVs, and exporting data from Salesforce.

Watch for: the load-performance toolkit. Bulk API 2.0 batches large files and runs them in parallel for you, and performs PK chunking automatically on queries, where the original Bulk API makes you configure both. Before a large load, defer sharing calculation suspends group membership and sharing rule calculation until the load is done. On export, the Data Export Service can be run manually every 7 days (weekly) or 29 days (monthly), depending on edition.

A Six-to-Ten-Week Study Plan

1. Read the exam guide and score yourself. Rate each objective from 1 to 3. Data modelers often find LDV and migration the weak spots, and together they're 35% of the exam.
2. Take a baseline. Try the 18 free Platform Data Architect practice questions to see the scenario style and find your weak sections.
3. Work the official trailmix. The credential page links Architect Journey: Data Architecture and Management, and the guide recommends the Build Data Architect Expertise (ARC201) course if you want instructor-led training.
4. Read the LDV and record-access guides end to end. Best Practices for Deployments with Large Data Volumes and Designing Record Access for Enterprise Scale are the source of most skew, indexing, skinny-table and deferred-sharing questions. Make your own list of every number they mention.
5. Build in a free Developer Edition org. Define a custom big object, set data classification on five fields, create a matching rule and duplicate rule, and load 50,000 test records with Data Loader using Bulk API 2.0.
6. Practice under time. Book the exam when you're comfortably above 58% and can say why each wrong design fails at scale. Our guide to answering Salesforce scenario questions helps with the "choose 2" format.

Platform Data Architect Practice Exam on Certifyforce

Certifyforce has a practice bank for this exam, written to the official outline and checked against Salesforce Help, the Salesforce Developers guides and Trailhead. It isn't affiliated with Salesforce, and it contains no recalled exam questions or brain dumps.

• The Platform Data Architect practice exam: 300+ reviewed questions across all six sections, with timed 60-question exams weighted like the real one, section practice, and an explanation and source for every answer. The full pack is launching soon.
• 18 free practice questions with answers, explanations and the official documentation each was checked against, no account needed.

Frequently Asked Questions

How many questions are on the Platform Data Architect exam?

60 scored multiple-choice/multiple-select questions plus up to 5 unscored ones, in 105 minutes, with a 58% passing score, per the exam guide.

How much does the Platform Data Architect exam cost?

US$400 (JPY 60,000) plus applicable taxes, and a retake costs US$200 (JPY 30,000), per the exam guide.

Does Platform Data Architect have a prerequisite?

No. The guide lists no prerequisite, though it recommends Platform App Builder, Platform Developer and Platform Developer II first.

Is Platform Data Architect required for Application Architect?

Yes. Application Architect is granted automatically when you hold Platform App Builder, Platform Data Architect, Platform Developer and Platform Sharing and Visibility Architect.

Does the exam cover Data 360?

The outline aligns to Spring '23 and doesn't name Data 360 (formerly Data Cloud). Its MDM and single-view objectives cover concepts Data 360 implements, but the exam is about core platform data architecture. Data 360 has its own consultant certification.

How do I keep the Platform Data Architect certification?

Complete the Platform Data Architect maintenance badge on Trailhead once a year. Missing the due date expires the certification.