Plat-Arch-204 Salesforce Certified Platform Integration Architect Exam Topics and Questions
These Salesforce Certified Platform Integration Architect (Plat-Arch-204) exam topics are organized according to official exam domains to help candidates quickly verify coverage and focus on assessment rather than theory. Each domain is paired with topic-wise Plat-Arch-204 sample questions that reflect how objectives are tested in the actual exam. This structure enables efficient review, targeted self-assessment, and rapid identification of weak areas when preparing for the Salesforce Certified Platform Integration Architect certification exam.
Let's Practice Free Salesforce Plat-Arch-204 Questions Aligned with Official Exam Topics
This topic opens the exam by forcing you to map what already exists before you design anything new. You need to read a requirements statement, identify every system in play, and spot the technical boundaries that will constrain your options. The cost comes from misreading the landscape: miss a legacy protocol constraint or an undocumented authentication scheme and your proposed integration fails before it starts. How Evaluate the Current System Landscape is tested At 8 per cent, this topic delivers...
This topic opens the exam by forcing you to map what already exists before you design anything new. You need to read a requirements statement, identify every system in play, and spot the technical boundaries that will constrain your options. The cost comes from misreading the landscape: miss a legacy protocol constraint or an undocumented authentication scheme and your proposed integration fails before it starts.
How Evaluate the Current System Landscape is tested
At 8 percent, this topic delivers around six items. Each one presents a scenario with multiple systems and asks you to identify what limits your choices. You will see requirements that conflict with existing protocols, authentication schemes that do not align, or authorization boundaries that block a proposed data flow. The trap is choosing the solution that works in a vacuum but ignores a stated constraint. Marks are lost when candidates pick the technically elegant answer over the one that respects the landscape as described. Read every sentence in the scenario: the constraint is always explicit, and the wrong answers assume it away.
The practice test gives you volume across all six topics, so you can verify coverage before the exam window opens. The PDF and timed mode both come with the same purchase, and both are available in demo form first.
The question below turns on identifying which part of the landscape creates the binding constraint.
Northern Trail Outfitters requires an integration to be set up between one of its Salesforce orgs and an External Data Source using Salesforce Connect. The External Data Source supports Open Data Protocol. Which configuration should an integration architect recommend be implemented in order to secure requests coming from Salesforce?
Once the landscape is clear, the exam shifts to what the business actually needs. This topic tests whether you can separate functional demands from performance expectations, classify data by sensitivity, and recognize when growth or regulation changes your integration choices. Candidates lose time here by treating every requirement as equal weight. β Functional versus non-functional requirements You must distinguish what the integration does from how well it does it. Functional requirements describe data movement, transformation, or synchronization. Non-functional requirements cover...
Once the landscape is clear, the exam shifts to what the business actually needs. This topic tests whether you can separate functional demands from performance expectations, classify data by sensitivity, and recognize when growth or regulation changes your integration choices. Candidates lose time here by treating every requirement as equal weight.
β Functional versus non-functional requirements
You must distinguish what the integration does from how well it does it. Functional requirements describe data movement, transformation, or synchronization. Non-functional requirements cover response time, throughput, availability, and error tolerance. The exam will present a use case and ask which category a given requirement belongs to, or it will ask what is missing. The trap is conflating the two: a requirement for real-time updates is non-functional, but a requirement to update the Account object is functional. If you cannot classify a requirement correctly, you cannot prioritize it or choose the right pattern to satisfy it.
β Data classification and CRM success factors
Data classification drives security and compliance decisions. Confidential data requires encryption and restricted access; public data does not. The exam gives you a use case and asks you to classify fields or records. CRM success factors are the business outcomes that justify the integration: user adoption, data quality, process efficiency. You need to recognize when a proposed integration undermines one of these factors, such as a batch process that delays updates and erodes trust in the system. The exam tests whether you can identify which success factor is at risk when a requirement changes. Growth and regulatory factors appear when scaling or compliance constraints force you to choose one integration approach over another. A solution that works for a hundred records per hour may fail at ten thousand, and a solution that satisfies domestic law may violate cross-border rules.
How Evaluate Business Needs is tested
Expect around eight items from this topic. Each one gives you a scenario and asks you to identify what kind of requirement you are reading, how sensitive the data is, or which business factor is in play. The wrong answers either misclassify a requirement or ignore a stated constraint. You will see options that treat a performance need as a functional one, or that classify customer payment data as public. Marks are lost when candidates skim the scenario and miss the word that changes the classification. Read the use case twice: the first pass identifies the domain, the second pass identifies the category. If the scenario mentions regulation, growth, or user adoption, that word is load-bearing.
Seeing item style before you pay helps you decide whether the bank matches your gaps. The demo covers every topic, so you can check phrasing and difficulty without committing.
The sample tests whether you can distinguish a non-functional requirement from a functional one under time pressure.
Which Web Services Description Language (WSDL) should an architect consider when creating an integration that might be used for more than one Salesforce org and different metadata?
This is the heaviest topic in the first half of the exam, and it tests your ability to convert business language into technical specifications. You take a use case, a diagram, or a process description and extract the integration requirements that will satisfy it. The cost is in the translation: miss a constraint or misread a performance threshold and your requirement set is incomplete. β System inventory and pattern recognition You receive a landscape diagram showing multiple systems and their...
This is the heaviest topic in the first half of the exam, and it tests your ability to convert business language into technical specifications. You take a use case, a diagram, or a process description and extract the integration requirements that will satisfy it. The cost is in the translation: miss a constraint or misread a performance threshold and your requirement set is incomplete.
β System inventory and pattern recognition
You receive a landscape diagram showing multiple systems and their connections. Your job is to list every system, identify the integration patterns in use, and recognize where patterns conflict or overlap. The exam expects you to name the pattern, whether request-reply, fire-and-forget, batch, or event-driven, and to spot when a diagram shows two patterns trying to solve the same problem. The trap is assuming a pattern from partial information. If the diagram does not show message acknowledgment, do not infer synchronous behavior. Inventory means accounting for every node and every link, because the one you skip is the one the question turns on.
β Constraints from process and system design
Business processes impose timing, sequencing, and approval constraints that limit integration options. A process that requires manual approval before data moves rules out fully automated synchronization. A system that batches updates overnight rules out real-time queries. The exam gives you a process map or a system description and asks what constraint it creates. You lose marks by choosing a solution that ignores the constraint, such as proposing a synchronous call to a system that only accepts batch files. Read the process end to end and identify every handoff, every wait state, and every approval gate. Those are your constraints.
β Security, authentication, and authorization requirements
Every integration crosses a trust boundary, and this sub-topic tests whether you know what that boundary demands. Authentication proves identity; authorization grants access. The exam describes a use case and asks what security mechanism is required. You need to distinguish between API keys, OAuth flows, SAML assertions, and mutual TLS, and you need to know when each applies. The trap is choosing the most secure option when the scenario requires the most compatible one, or choosing the simplest option when compliance mandates stronger controls. If the use case mentions external users, third-party systems, or regulated data, security requirements tighten.
β Performance thresholds and solution fit
Performance needs are stated as volume, response time, or latency. A requirement for sub-second response rules out batch. A requirement for ten thousand records per hour rules out individual API calls. The exam gives you numbers and asks which solution meets them. You must know governor limits, API call limits, and the performance envelope of each integration pattern. The wrong answers either ignore the numbers or propose a pattern that cannot scale to the stated volume. If the scenario gives you a number, that number will determine the correct answer. Calculate whether the proposed solution can handle the load, and if it cannot, eliminate it.
How Translate Needs to Integration Requirements is tested
This topic delivers around sixteen items, more than any other. Each one presents a scenario, a diagram, or a process and asks you to extract a specific requirement. You will see questions that ask which pattern is in use, which constraint blocks a proposed solution, which security mechanism applies, or which performance threshold rules out an option. The trap is reading the scenario too quickly and missing the detail that changes the answer. Candidates lose marks by choosing solutions that work in general but fail the specific constraint stated in the stem. Every scenario contains one fact that eliminates three of the four options. Find that fact first, then evaluate the remaining choice.
Volume matters most for this topic because it carries the highest weight. Working through enough items reveals the pattern of how constraints are described and how wrong answers are constructed.
The sample below asks you to identify the performance requirement that determines which integration pattern applies.
This is the single heaviest topic on the exam, and it tests your ability to choose the right integration pattern, API, and component set for a given use case. You must weigh trade-offs, recognize limitations, and select the standards and security mechanisms that fit the technical constraints. Candidates lose marks by choosing the pattern they know best rather than the one the scenario demands. β Pattern selection and component definition You are given a use case and asked which integration...
This is the single heaviest topic on the exam, and it tests your ability to choose the right integration pattern, API, and component set for a given use case. You must weigh trade-offs, recognize limitations, and select the standards and security mechanisms that fit the technical constraints. Candidates lose marks by choosing the pattern they know best rather than the one the scenario demands.
β Pattern selection and component definition
You are given a use case and asked which integration pattern applies: remote process invocation, fire-and-forget, batch data synchronization, remote call-in, or UI update. Each pattern has a signature. Remote process invocation requires immediate response. Fire-and-forget tolerates delay. Batch handles volume but not urgency. The exam expects you to match the pattern to the requirement, then define the components that implement it, such as middleware, message queues, or platform events. The trap is choosing a pattern because it is familiar or modern, not because it fits. If the use case states that the external system must respond before Salesforce proceeds, only one pattern applies. If it states that Salesforce must not wait, a different pattern applies. Read the timing requirement first.
β Trade-offs, limitations, and constraints
Every integration pattern has a cost. Synchronous calls block the user. Asynchronous calls complicate error handling. Batch introduces latency. The exam describes a solution and asks what trade-off it creates, or it asks which limitation prevents a pattern from working. You need to know governor limits, API call limits, timeout thresholds, and the failure modes of each pattern. The wrong answers ignore a stated constraint or propose a solution that exceeds a governor limit. If the scenario mentions that the external system takes five seconds to respond, any solution that assumes a two-second timeout is wrong. If it mentions fifty thousand records, any solution that hits the batch Apex row limit is wrong. Constraints are explicit; limitations are intrinsic to the pattern.
β API selection for Salesforce integrations
Salesforce offers REST, SOAP, Bulk, Streaming, Metadata, and Tooling APIs. Each serves a different use case. REST suits lightweight, stateless calls. SOAP suits enterprise systems that require WSDL contracts. Bulk handles large data volumes. Streaming pushes changes to subscribers. The exam gives you a technical requirement and asks which API applies. You must know the governor limits and the data shape each API expects. The trap is choosing REST because it is common when the scenario requires Bulk because of volume, or choosing SOAP because the external system is enterprise-grade when REST would satisfy the requirement with less overhead. If the use case states tens of thousands of records, Bulk is the answer. If it states near-real-time notification of changes, Streaming is the answer.
β Standards, components, techniques, and security
This sub-topic tests your knowledge of OAuth flows, SAML, JSON Web Tokens, middleware platforms, API gateways, and encryption standards. The exam describes a technical constraint or driver and asks which standard or component applies. You need to distinguish OAuth 2.0 flows, know when to use named credentials versus custom settings, and recognize when mutual TLS is required. The wrong answers either propose a technique that does not satisfy the security requirement or propose a component that introduces unnecessary complexity. If the scenario mentions a third-party mobile app, OAuth user-agent flow applies. If it mentions server-to-server integration with no user context, OAuth client credentials or JWT bearer flow applies. If it mentions encryption in transit for regulated data, TLS is required and the version matters.
How Design Integration Solutions is tested
Expect around twenty items from this topic, more than any other. Each one gives you a use case with technical requirements and asks you to choose the pattern, API, component, or security mechanism. The wrong answers are plausible but fail on one dimension: they exceed a limit, ignore a constraint, or introduce a trade-off the scenario cannot accept. Candidates lose marks by skimming the requirements and picking the first option that sounds reasonable. The correct answer satisfies every constraint stated in the scenario. The wrong answers satisfy most but not all. Read the scenario twice: once for the domain, once for the numbers and the constraints. If the scenario gives you a response time, a record count, or a security requirement, that detail determines the answer.
Timed practice helps you build speed on the heaviest topic without sacrificing accuracy. Working under exam conditions reveals whether you are reading scenarios carefully enough to catch the constraint that eliminates three options.
The question below asks you to select the API that satisfies both the volume and the timing requirement.
Design is complete; now you build it. This topic tests the practical considerations that determine whether an integration survives production: API design, outbound call strategy, scalability, error handling, security implementation, and resilience during system updates. The cost is in the details that break under load or fail during deployment. β API provider and consumer considerations When Salesforce provides an API, you must consider versioning, authentication, rate limiting, and response format. When Salesforce consumes an API, you must consider callout limits,...
Design is complete; now you build it. This topic tests the practical considerations that determine whether an integration survives production: API design, outbound call strategy, scalability, error handling, security implementation, and resilience during system updates. The cost is in the details that break under load or fail during deployment.
β API provider and consumer considerations
When Salesforce provides an API, you must consider versioning, authentication, rate limiting, and response format. When Salesforce consumes an API, you must consider callout limits, timeout handling, and certificate management. The exam gives you a use case and asks what you must account for in the implementation. You need to know that named credentials simplify authentication but require setup, that remote site settings are mandatory for outbound calls, and that API versioning prevents breaking changes from disrupting integrations. The trap is assuming the default configuration suffices. If the external system requires mutual TLS, you must configure a certificate. If the external API has a rate limit, you must implement retry logic or queuing.
β Outbound calls and scalability
Making an outbound call from Salesforce means choosing between a synchronous callout, an asynchronous method, or a platform event. Synchronous calls block the transaction and count against governor limits. Asynchronous calls via future methods or queueable Apex avoid blocking but complicate error handling. The exam asks which option applies given the use case. Scalability considerations include bulkification, avoiding callouts in loops, and respecting API call limits. The wrong answers propose solutions that work for one record but fail at scale. If the scenario mentions a trigger that fires on bulk insert, any solution that makes a callout per record is wrong. You must aggregate calls or move them to an asynchronous context.
β Error handling across integration options
Error handling varies by pattern. Synchronous integrations can return errors immediately, but you must decide whether to roll back the transaction or log the error and continue. Asynchronous integrations require retry logic, dead-letter queues, or manual intervention. Batch integrations require partial success handling and error logs that identify which records failed. The exam describes an integration option and asks how errors should be handled. You need to know when to use platform events for retry, when to use custom objects for logging, and when to fail fast versus fail gracefully. The trap is proposing error handling that introduces more complexity than the use case justifies, or proposing none at all.
β Security and resilience
Security for inbound integrations means validating requests, enforcing authentication, and applying field-level and object-level permissions. Security for outbound integrations means encrypting credentials, using named credentials, and avoiding hardcoded secrets. Resilience means building integrations that tolerate system updates, API version changes, and temporary outages. The exam asks what you must do to secure an integration or to ensure it survives a Salesforce release. You need to know that platform events decouple publishers from subscribers, that API versioning isolates your integration from breaking changes, and that circuit breaker patterns prevent cascading failures. The wrong answers either ignore a security risk or assume the integration will never need to change.
How Build Solution is tested
This topic delivers around seventeen items. Each one presents a technical scenario and asks what consideration applies when building the integration. You will see questions about callout limits, error handling strategies, security mechanisms, and resilience patterns. The wrong answers are incomplete: they address part of the requirement but ignore a governor limit, a security risk, or a failure mode. Candidates lose marks by choosing the answer that works in a happy-path scenario but breaks under load or during an error condition. Read the scenario for volume, for failure conditions, and for security requirements. If the scenario mentions bulk operations, governor limits apply. If it mentions external users, authentication and authorization apply. If it mentions system updates, resilience applies.
Checking a weak area by working through items is faster than re-reading the documentation. The practice test isolates your gaps so you can focus study time where it matters.
The sample turns on choosing the error handling approach that satisfies both the reliability requirement and the governor limits.
The final topic tests what happens after the integration goes live. You must identify what to monitor, how to handle failures, and what to report. At 8 per cent this is the lightest topic, but candidates still lose marks by ignoring operational concerns that determine whether an integration remains reliable over time. How Maintain Integration is tested Expect around six items. Each one describes an integration in production and asks what monitoring, error handling, or reporting is required. You will...
The final topic tests what happens after the integration goes live. You must identify what to monitor, how to handle failures, and what to report. At 8 percent this is the lightest topic, but candidates still lose marks by ignoring operational concerns that determine whether an integration remains reliable over time.
How Maintain Integration is tested
Expect around six items. Each one describes an integration in production and asks what monitoring, error handling, or reporting is required. You will see scenarios that ask which metric indicates performance degradation, how to escalate a failed batch job, or what report stakeholders need to verify SLA compliance. The wrong answers either monitor the wrong metric, propose escalation procedures that introduce delay, or generate reports that do not answer the business question. Candidates lose marks by choosing answers that sound operationally mature but do not address the specific need stated in the scenario. If the use case mentions SLA compliance, the correct answer involves tracking response time or uptime. If it mentions failed records, the correct answer involves logging and retry counts. Read the scenario for what must be measured, not what could be measured.
Coverage across all six topics ensures no gap costs you marks. The bank includes items from this final topic so you can verify you understand operational concerns as well as design and build.
The sample below asks which monitoring approach detects the failure mode described in the scenario.
Ready to Start Practicing?
Access all questions and start your exam preparation journey
Upgrade to Full Plat-Arch-204 Exam Questions π