# Free SIA practice questions

> 10 free SIA practice questions with the correct answer and an explanation.

Page: https://devopscodelab.com/exams/salesforce-integration-architect/free-practice-questions/
Practice exams: https://devopscodelab.com/exams/salesforce-integration-architect/

### Question 1 of 10

An architect finds that an existing integration authenticates as a named employee's user account, which has the System Administrator profile. What should the architect recommend?

- A. Use a Customer Community license for the integration user because it costs less than a full license
- B. A dedicated integration user with an API-only profile and permission sets granting only the needed access
- C. Create a second System Administrator user so the integration no longer depends on the employee leaving
- D. Keep the employee's account but enable multi-factor authentication so that the API calls are better protected

**Answer:** B

**Explanation:** Integrations should run as a dedicated integration user, such as one with the Salesforce Integration license and Minimum Access - API Only Integrations profile, with permission sets for least privilege. Adding MFA to an employee account does not fix shared identity or excessive privilege, and API logins cannot complete interactive MFA. A second administrator still grants far more than needed. Community licenses are for external users and do not provide the required internal API access.

### Question 2 of 10

During discovery for a new integration, which question BEST helps the architect determine the system of record for customer billing addresses?

- A. How many API calls per day does the billing application receive from all of its currently connected systems?
- B. Which application is authorized to create and change billing addresses, and which systems only consume them?
- C. Which middleware product has the company already licensed for its other enterprise application integrations?
- D. What response time do sales users expect when they open an Account page that shows the billing address?

**Answer:** B

**Explanation:** The system of record is the application that owns the data, meaning it is the authoritative place where the data is created and maintained; asking who can create and change the address versus who only reads it identifies that owner. API call volume and licensed middleware are useful for sizing and tooling but do not establish ownership. Expected page response time is a non-functional requirement, not a data-ownership question.

### Question 3 of 10

A pharmaceutical company wants a lab system to be notified when a Clinical Trial record's status changes to Approved. The lab system can host an HTTPS endpoint, but it requires the notification to be signed and the payload to contain only the trial ID and status. Volume is about 50 changes per day. Which option BEST balances simplicity and the stated needs?

- A. Change Data Capture on Clinical Trial with the lab system subscribing over the Pub/Sub API using a managed subscription
- B. An outbound message containing the trial ID and status, with the lab system validating the session ID
- C. A record-triggered flow using an HTTP Callout action through a Named Credential to send the two fields
- D. A nightly batch export of approved trials sent as a CSV file to the lab system by SFTP

**Answer:** C

**Explanation:** A flow HTTP Callout through a Named Credential sends a minimal, custom payload with configured authentication at low volume and without heavy infrastructure. Change Data Capture requires the lab system to build a subscriber and sends change event headers rather than a simple notification. Outbound messages use a fixed SOAP envelope rather than the requested minimal payload. A nightly batch does not notify when the status changes.

### Question 4 of 10

A logistics company needs Salesforce users to see a shipment's current status from a carrier system when they open a Shipment record. The data changes every few minutes, must not be stored in Salesforce, and the carrier exposes an OData 4.0 endpoint. Which approach meets these requirements with the LEAST custom code?

- A. Salesforce Connect with the OData 4.0 adapter, surfacing the shipments as an external object related to the record
- B. A nightly Bulk API 2.0 load from the carrier into Salesforce, with a formula field that shows the most recent status
- C. A scheduled Apex job that copies carrier statuses into a custom object every five minutes for display on the page
- D. A Lightning web component that calls the carrier endpoint through an Apex controller and caches the response in a custom field

**Answer:** A

**Explanation:** Salesforce Connect with the OData 4.0 adapter implements the data virtualization pattern: records are read live from the carrier and are never stored in Salesforce, and no code is needed. The scheduled job and nightly Bulk load both copy data into Salesforce and would be stale between runs. The Lightning web component needs custom Apex and caching the response in a field stores the data, which violates the requirement.

### Question 5 of 10

An insurer must send each new Claim record to a fraud-scoring service. Users should not wait for the score, the service can take up to 40 seconds to respond, and the architect wants guaranteed delivery with the ability to replay missed messages if the middleware is down for a few hours. Which design BEST meets these requirements?

- A. Expose an Apex REST resource that the middleware polls every minute for any claims created since its last successful request
- B. Call the fraud service synchronously from an Apex before-insert trigger and roll back the claim if the call times out
- C. Use an @future method with callout=true from the claim trigger and log every failed call to a custom object for manual review
- D. Publish a platform event after commit from a flow and have the middleware subscribe via the Pub/Sub API using replay IDs

**Answer:** D

**Explanation:** A platform event published after commit decouples the user transaction (fire-and-forget), and the event bus retains events so a subscriber using the Pub/Sub API can resume from its last replay ID after an outage. Callouts are not allowed in a before-insert trigger and would make the user wait anyway. An @future callout has no built-in replay and failures need manual handling. Polling an Apex REST resource adds latency and API consumption and requires the middleware to track state itself.

### Question 6 of 10 (choose 2)

A media company streams subscription changes from Salesforce to a downstream analytics platform. The platform is occasionally offline for up to a day, events must not be lost, and the company wants to minimize custom code inside Salesforce. Which two design choices support these requirements? (Choose two.)

- A. Use Change Data Capture on the Subscription object so change events are generated without custom publishing code
- B. Rely on the event bus retaining events for 30 days so the subscriber can catch up after any outage
- C. Have the subscriber persist the last processed replay ID and resume from it after reconnecting
- D. Use an outbound message so that Salesforce stores undelivered notifications until the platform returns
- E. Publish events with the Publish Immediately behavior so they reach the bus even if the transaction rolls back

**Answer:** A, C

**Explanation:** Change Data Capture generates change events declaratively, and a subscriber that stores its last replay ID can resume where it left off after reconnecting, within the retention window. High-volume events and change events are retained for 72 hours, not 30 days, so a day-long outage is covered but not a month. Publish Immediately can deliver events for transactions that later roll back, which risks sending changes that never happened. Outbound messages retry for a limited period and are not a streaming replay mechanism for an analytics consumer.

### Question 7 of 10

A bank wants its mobile app to let customers update their contact preferences, stored in Salesforce, while the app's traffic must pass through the bank's API gateway for throttling, logging, and threat protection. Customers authenticate with the bank's own identity provider. Which design is MOST appropriate?

- A. Each mobile app instance calls the Salesforce REST API directly using a shared username and password embedded in the app
- B. The app embeds a Salesforce login page so customers create Salesforce credentials alongside bank ones
- C. The gateway exposes an experience API and calls Salesforce with a server-side integration identity
- D. The app writes preferences to a file that a nightly batch uploads to Salesforce through Data Loader

**Answer:** C

**Explanation:** Routing calls through the bank's gateway as an experience API keeps throttling, logging, and threat protection in place, and the gateway authenticates to Salesforce server-to-server (for example with the JWT bearer flow) while customers authenticate with the bank's identity provider. Embedding shared credentials in a mobile app is insecure and bypasses the gateway. A second set of Salesforce credentials duplicates identity management. A nightly file load does not provide an interactive update.

### Question 8 of 10

A Queueable Apex job sends invoices to an external billing API. When the external system returns HTTP 503, the business wants the job retried up to three times and a support case created if all attempts fail, even when the job ends with an unhandled exception. Which Apex feature is MOST suitable for implementing this behavior?

- A. A Transaction Finalizer attached with System.attachFinalizer that inspects the result and re-enqueues the job or creates the case
- B. A Batch Apex class implementing Database.Stateful whose finish method counts the failed invoices and resubmits the full batch
- C. A scheduled Apex job that runs every hour and resubmits any invoice whose status field is not yet set to Sent
- D. A try-catch block around the callout that calls System.enqueueJob in the catch block and creates the case after the third failure

**Answer:** A

**Explanation:** A Transaction Finalizer runs after the Queueable finishes whether it succeeded or failed with an uncaught exception, so it can check the result, re-enqueue the job with a retry count, and create a case after the last attempt. A try-catch cannot catch every failure, such as uncatchable limit exceptions, so it does not guarantee the case is created. A stateful Batch job is a heavier construct that retries whole batches rather than the single failed job. An hourly scheduled job adds delay and does not track individual retry counts.

### Question 9 of 10 (choose 2)

An external client application must call Salesforce REST APIs on behalf of individual users who log in through a browser. The security team requires that no user passwords pass through the client application. Which two OAuth 2.0 flows are appropriate? (Choose two.)

- A. Client credentials flow with a single shared run-as user
- B. JWT bearer flow using a certificate stored in the client
- C. Web server flow (authorization code) with PKCE
- D. User-agent flow for a browser-based single-page application
- E. Username-password flow with the user's security token

**Answer:** C, D

**Explanation:** The web server flow with PKCE and the user-agent flow redirect users to Salesforce to log in, so the client never handles their passwords, and both act on behalf of the individual user. The username-password flow requires the client to collect the password. The JWT bearer flow and the client credentials flow are server-to-server flows that do not involve interactive user login, and the client credentials flow runs as one shared user rather than each individual.

### Question 10 of 10

An outbound integration that posts orders to a fulfillment API began failing after the partner rotated its TLS certificate to a new certificate authority. Which item should the Salesforce team check FIRST?

- A. Whether the new CA is trusted by Salesforce or must be imported in Certificate and Key Management
- B. Whether the Order object's field-level security still allows the integration user to read fields
- C. Whether the integration user's API Enabled permission was removed from its assigned permission set
- D. Whether the org's daily API request allocation was exceeded by the order integration

**Answer:** A

**Explanation:** Outbound callouts require Salesforce to trust the server certificate's chain, so a change of issuing CA can break the TLS handshake; the team should confirm the CA is trusted or import the certificate. API Enabled and the daily API allocation govern inbound API calls into Salesforce, not outbound callouts. Field-level security would cause data errors, not a sudden TLS failure that coincides with a certificate change.

Ready for a timed run? Take the free 15-question SIA mini exam: https://devopscodelab.com/app/exams/salesforce-integration-architect/mini/

DevOpsCode Lab is not affiliated with Salesforce. SIA and related names are trademarks of their owners. We write every question ourselves; these are not real exam questions.
