Salesforce Security Assessment Permissions, Sharing, and Shield Gaps

A Salesforce security assessment tests whether people and connected systems can reach only the data they need. It checks who can log in, what each user can do, which records they can see, and whether sensitive activity leaves an audit trail. The goal is to find access that has grown beyond its business purpose and close it without stopping legitimate work. A Salesforce security assessment is strongest when business owners approve the expected access before any test begins. Teams that want an experienced partner can look at VALiNTRY360’s Salesforce consulting.

The problem often develops quietly. A sales manager gets temporary admin access for a project. A contractor keeps an old permission set. A sharing rule created for 1 region reaches a second team after a reorganization. Meanwhile, encryption and event logs may be switched on without anyone checking what they actually protect. Each setting can look reasonable in isolation. Combined, they may expose customer records or leave investigators unable to explain a data export.

A useful review follows the path a real user takes through Salesforce. It starts with sign-in, moves through system and object permissions, tests record sharing, and finishes with monitoring and evidence. It also checks custom code because a secure page layout doesn’t prove that an API or an Apex class respects the same boundaries. This guide explains the tests, common mistakes, and decisions needed to make the findings useful to both administrators and security owners.

TL:DR

A Salesforce security assessment should test effective user access rather than count settings that appear secure. Check MFA and privileged accounts first. Trace access granted through profiles, permission sets, groups, object rights, field rights, and every route to a record. Test integrations and public sites separately. Then examine Salesforce Shield features for usable logs, encryption coverage, and field history retention. Record evidence for each gap, rank it by the data exposed and how easily it could be misused, and retest after changes. A good result is a documented access decision that the business can explain and maintain.

What a Salesforce security assessment should cover

The main deliverable is a tested account of who can do what. A settings inventory helps, but a list of checkboxes can’t tell whether a user can export customer financial fields or view a confidential opportunity. Define a small set of high-value records for testing. Use fictitious or approved masked data where possible. The assessor should document how each test account was created and remove it after testing. For each important business role, an assessor should compare expected access with observed access in a controlled test. That comparison reveals both excess privileges and barriers that push staff toward unsafe workarounds.

Start by agreeing which Salesforce environments and connected products are in scope. A production Sales Cloud org might share data with Service Cloud users, an Experience Cloud portal, a marketing application, or a warehouse integration. Each path can change who receives information. Include sandboxes if they contain production data or allow privileged access. Don’t assume security in a test environment is harmless just because its business users aren’t active.

Salesforce offers a useful starting screen. Its Security Health Check compares selected settings with a baseline and highlights risks in password and session controls. The score doesn’t establish that record sharing is correct or that a sensitive field stays hidden from a report. Treat it as one input to a wider assessment, with its baseline and date recorded for later comparison.

A sound evidence pack names the finding, affected users or services, access path, business impact, owner, and repeatable test. Record the expected behavior before a change is made. A finding such as “sales users can read payroll records” is far more useful than “sharing is weak.” It gives the owner a clear failure to reproduce and a result to verify.

Start with identity, MFA, and privileged sessions

Most access reviews begin too far inside the application. If an attacker can sign in as an administrator, later record rules may offer little protection. List active users, inactive users with integrations tied to their identity, emergency accounts, delegated administrators, and service accounts. Compare each person with HR or contract records. Flag accounts whose owners have left, whose login method isn’t documented, or whose role changed without an access review.

Test authentication paths

Check direct Salesforce login, single sign-on, mobile access, and any approved recovery route. Salesforce’s MFA requirements cover employee UI logins through direct sign-in and SSO. Since summer 2026, privileged UI users must use phishing-resistant MFA, such as passkeys or FIDO2 security keys; SSO must convey recognized authentication signals. API-only integration logins aren’t subject to that UI requirement, but a person signing into an integration account through the UI is. Verify each route rather than relying on an identity-provider policy screenshot.

Inspect session timeout settings, login restrictions, trusted network assumptions, and processes for resetting authenticators. A strict session policy won’t help if a shared emergency account has a password known to several people. Emergency access needs an owner, limited purpose, clear approval, and a way to review its use after the incident. Revoke sessions when access is removed where the situation requires it.

Separate administrators from daily work

Ask why each user has powerful permissions such as Modify All Data, View All Data, or Manage Users. These rights can bypass ordinary record restrictions or change the rules themselves. Give system administrators separate administrative identities where the organization’s operating model supports that split. Avoid using permanent administrator access for routine reporting or case handling. An access register should state who approved each exception and when it must be reviewed.

This identity pass sets the floor for the Salesforce security assessment. Once sign-in and privileged roles are understood, the team can test what lower-privilege users can access without confusing normal permissions with an administrator’s broad rights.

Audit profiles, permission sets, and permission groups

Salesforce permissions are cumulative across several places, as the permissions model explains. A user has a profile, may hold multiple permission sets, and may receive more permissions through groups. A restrictive profile alone doesn’t mean the person has restricted access. A Salesforce permission audit has to examine the full assignment stack. The reviewer needs the effective result across all assignments, including old project permissions that never expired.

Salesforce’s user access guidance describes profiles as a home for default settings and recommends permission sets and groups for granting access. Start with job-based access rather than 1 permission set per individual. Compare what an inside sales representative, customer support agent, and finance analyst genuinely need. Differences should be explained through duties, not job titles alone.

A Salesforce permission audit should flag broad grants, duplicate assignments, inactive permission sets, and special permissions with no business owner. Review system permissions alongside object permissions. Access to the API, report exports, user management, and setup functions can create greater exposure than a single object-level Read grant. Managed packages also deserve attention because a package permission set can change as the app is updated.

Don’t mistake muting for a deny rule

A common error is to mute a permission in a group and assume the user can no longer exercise it. Salesforce’s muting guidance explains that muting affects only its permission set group. The user may still receive the same permission from a separate assignment or their profile. The assessor must trace every grant. Removing an unneeded direct assignment may be safer than adding another layer of exceptions. Inspect effective permissions and all assignment sources for every in-scope identity, using User Access Summary and assignment exports. Sample representative users only for controlled allow-and-deny tests, then test known outliers separately. A role baseline helps spot repeated design problems.

Compare each role’s intended permission package with actual assignments. Keep the original role matrix as the reference for later checks. Review time-limited permissions on a schedule and ensure changes to team membership trigger another check. The purpose is to make access predictable, so that a manager can explain why a user can perform a sensitive action without opening several unrelated configuration pages.

Verify object and field access against real tasks

Object permissions determine whether a user can create, read, edit, or delete records of a given type. Record sharing then controls which individual records are available. That distinction matters: making an opportunity private won’t remove a broad object permission, and giving someone access to a record won’t give them object Read access if it’s missing. Test both layers as a pair.

At this stage, build a small matrix for critical objects such as Accounts, Contacts, Opportunities, Cases, Contracts, and custom records containing financial or personal data. For each business role, record expected Create, Read, Edit, and Delete rights. Pay special attention to View All and Modify All at the object level. Those settings can defeat the intended limits of ordinary sharing for that object. Query permissions and related API operations need to match the same policy.

Field permissions add another boundary. In Salesforce’s field permissions guide, visibility and edit rights are managed separately for fields within an object. A field hidden from a page layout isn’t necessarily protected from reports or the API. Test confidential salary, customer identification, contract discount, and bank-related fields through more than 1 interface. A Salesforce permission audit that checks only page layouts may miss the most important exposure.

Test both allowed and forbidden actions

Choose a typical user for each role and attempt legitimate work first. Can the person create the case, update the required status, and read the fields needed to respond? Then attempt an action that policy forbids, such as exporting a sensitive report or editing a restricted discount field. Capture the result, expected outcome, and tested user. Negative tests must not use live personal data or cause destructive changes.

One department may need to view a sensitive field without changing it. Another may need to edit it but only for records it owns. Distinguish these cases before applying fixes. A broad revoke may improve the report’s appearance while causing staff to share spreadsheets or borrow credentials. The right repair preserves the approved task and closes the unapproved path.

Review org-wide defaults and record sharing paths

Salesforce sharing settings answer a different question from object permissions: which specific records can a user reach? Begin with Salesforce’s organization-wide defaults for each important object. When the business needs limited record visibility, a restrictive default usually makes exceptions easier to account for. Public Read/Write may be valid for shared reference data but can expose customer information if selected simply to stop access complaints.

Walk through all ways a user can gain record access. Ownership may grant it. A role manager may inherit it. Criteria-based or owner-based sharing rules can extend it to groups. Manual sharing, account teams, territory management, and related-record access can create additional routes. Salesforce’s account access explanation shows why someone may see an account they were never directly assigned. An assessment should identify the exact route rather than guess based on the user’s department.

Use a realistic example: a regional seller should read only accounts in an assigned territory, while a service agent may need access to an account because of an open case. Test a record owned in another region. If the seller can see it, check sharing hierarchy or the relevant access detail before changing the role hierarchy. The access may come from an account team, an inherited permission, or a related record rather than the default you expected.

Review public groups whose membership includes roles or nested groups. Changes to group membership can silently expand visibility. Record who owns each sharing rule, what business requirement it serves, and what happens when that requirement ends. For a criteria-based rule, inspect the field values that make records eligible. A single incorrectly populated region field could share more records than the rule author intended. Salesforce sharing settings should be explained in business language: “regional sales leads can view open opportunities for their region,” not “Rule 14 grants Read.”

Examine role hierarchies and sharing exceptions

Role hierarchy reviews focus on inherited access. A role may reflect the reporting line when created, then become inaccurate after teams reorganize. A manager might inherit records from people no longer under their business authority. Compare the hierarchy with current reporting needs and note where access is broader than management responsibility requires.

Account and opportunity teams add another layer to Salesforce sharing settings. Review how members are added, what access level they receive, and who removes them when a deal changes hands. Territory assignments can persist after a sales region is redrawn. Test one record that changed ownership recently and another that moved between business units. They often reveal problems that a sample of stable records won’t.

Restriction rules can help narrow visibility in supported cases, but their scope is specific. Salesforce’s restriction-rule guidance lists custom objects, external objects, quotes, contracts, events, tasks, time sheets, and time sheet entries. Enterprise and Developer editions allow up to 2 active rules per object; Performance and Unlimited allow up to 5. Rules filter records a user could otherwise access. They don’t replace object or field permissions, and their behavior and feature coverage need testing before use.

Exceptions need an expiration plan. A temporary project group should have a documented end date. A manager-level exception should have a named approver. If a record requires special protection, test whether implicit account access or a privileged permission still defeats the intended restriction. A solid Salesforce security assessment examines the exception in the context of all other grants, then tests its removal in a sandbox or controlled process.

Check Experience Cloud, guests, and connected applications

External access deserves its own place in a Salesforce security assessment because customer and partner identities operate under a different business relationship from employees. Inventory active Experience Cloud sites, self-registration flows, public pages, and external user permission sets. Identify which objects a partner can see and whether records belong to the correct customer or partner account. Test users from 2 separate outside organizations to detect cross-account visibility.

Public guest permissions can leave serious Salesforce security gaps even where the employee access model is sound. Public sites can have a guest profile. Salesforce’s guest access guidance describes enforced private record defaults for guests, but older customizations still need examination. Check every exposed object and field, public form, file attachment, and custom component. Don’t assume a newer secure default corrected a configuration carried forward from years ago.

Connected applications and integrations may use OAuth, named credentials, API users, or other authorized paths. Build an inventory with owner, purpose, scopes, accessed objects, data destination, and authentication method. An integration that reads all account records to update 1 billing field should be questioned. Limit permissions to the process the connection performs, and remove connections that are no longer used.

For integrations, also check whether integration activity can be distinguished from employee actions. Shared service identities blur accountability. Review tokens and their rotation rules, logging coverage, and the response plan when a vendor contract ends. Test whether an old integration can still query records after it should have been retired. These are often high-impact Salesforce security gaps because they bypass the screens that ordinary user tests cover.

Test Apex, Flows, reports, and data exports

Custom automation deserves a separate pass in a Salesforce security assessment because it can weaken access boundaries even when the visible configuration looks correct. Review Apex classes, Lightning components, invocable methods, flows, and integration code that run with privileged context. Identify where business logic reads or changes sensitive records. A user-facing form may enforce field visibility while a server-side action retrieves more data than the user should see.

Apex behavior depends on API version and execution context. Salesforce’s Summer ’26 developer update describes user-mode database operations and `with sharing` defaults for classes compiled at API version 67.0. Triggers still run in system mode, so their sensitive work needs explicit checks in handler code. Classes at API version 66.0 or earlier need a separate review because they retain older system-mode defaults. Inspect each query, write, invocable method, and explicit `SYSTEM_MODE` operation. `with sharing` alone doesn’t enforce object or field permissions.

Follow the data beyond the screen

Assess reports and dashboards with sensitive fields. Test folder sharing, report export permissions, scheduled report recipients, and dashboards running as another user. Ask whether the intended person should receive the whole dataset or a narrower aggregate. A report sent to a broad distribution list can defeat careful record access design without changing the object itself.

Flows and automation need a specific review of running context, since Flow execution modes can enforce user access or run with elevated access. Confirm whether actions enforce the intended user permissions and what happens when they create or update records on someone’s behalf. Use test cases for a permitted operation, a forbidden operation, and a changed user role. Record the behavior after deployments because small code changes can reopen Salesforce security gaps that were fixed in configuration months earlier.

Assess Salesforce Shield Event Monitoring coverage

Salesforce Shield assessment work begins by separating features from evidence. Buying Shield or enabling an event feature doesn’t guarantee that someone receives an alert, can interpret it, or retains the log long enough for an investigation. Identify the event types important to your business: logins, API use, report exports, bulk changes, and access failures. Then test whether the data exists and whether a responsible team actually reviews it.

Salesforce’s Event Monitoring guidance explains how administrators configure streaming and stored real-time events. Availability and retention depend on the subscription and event type. Record which event families are enabled, the storage destination, who can read them, and what alert rule responds to a suspicious pattern. Don’t claim that every event is automatically captured forever.

Test an approved report export and confirm the event reaches its assigned owner with usable identity and time details. Salesforce’s 2026 Transaction Security update introduced a default `ReportEvent` policy for eligible Shield or Event Monitoring orgs without a matching existing condition-builder policy. It requires step-up MFA for UI report exports exceeding 10,000 rows; ReportEvent policy documentation explains the conditions. It doesn’t cover API exports. Confirm eligibility, whether the policy is enabled, and what logs or alerts reach the security team.

The practical test is how quickly the organization can answer “who exported this dataset?” If it cannot identify the user, source session, and event time from retained evidence, logging needs repair. Also restrict access to the logs themselves because they can reveal sensitive user behavior or system details.

Check platform encryption and key decisions

Encryption protects data under defined conditions, but it doesn’t replace permissions. An authorized user can still read data the application decrypts for them. Start by identifying which fields and files contain regulated or highly sensitive information, then ask which encryption method applies and whether the data can still be searched or used as the business requires.

Salesforce explains Shield Platform Encryption through tenant-controlled key material and supported methods. Field-level encryption and database-level protection don’t have identical behavior. Examine actual licensed features and supported fields in the org before stating that Shield covers an entire dataset. Review key ownership, creation, rotation, storage, and recovery procedures with the team responsible for continuity.

Check search and workflow effects

Probabilistic field encryption offers stronger resistance to pattern analysis but can restrict filtering and exact-match behavior. Salesforce’s encryption scheme comparison identifies where deterministic encryption supports matching. These choices have different security tradeoffs and need testing against the actual application. A Salesforce Shield assessment should list fields where encryption would interrupt matching, automation, duplicate checks, or reports. Test the exact use case with representative nonproduction data before a rollout. Don’t treat the availability of encryption as proof that it is appropriate for every field.

During encryption review, look for broad access to key administration and related settings. A sensitive key operation should require clear authority and leave a record. Document what staff should do if encryption or key changes affect applications. One of the harder Salesforce security gaps is an encryption project that appears finished while unexamined extracts, integrations, or other copies still contain the same sensitive information.

Review Field Audit Trail and evidence retention

Field history addresses a different need from event logs. Event logs can show actions such as an export or login. Field history can show changes to selected record values. For disputes about ownership, approval, or regulated fields, determine which changes the organization must reconstruct and whether the chosen history settings provide the needed detail.

Salesforce’s Field Audit Trail update says the feature no longer enforces the previous 10-year retention ceiling once Field Audit Trail is enabled. That doesn’t mean an organization should keep every record forever. Retention still needs to reflect contractual, legal, and operational requirements, including deletion rules. Confirm exactly which objects and fields are tracked and when collection started.

Test a permitted change to a selected field and confirm the before-and-after evidence can be retrieved by an approved user. Review changes performed by automation and integrations separately. If history must be exported for an audit, document the export format and who is allowed to extract it. Also check whether logs and field history use different retention policies and whether investigators know where each type of evidence lives.

A missing audit record can’t be created retroactively by buying an add-on after an incident. Mark coverage gaps honestly. If the organization changed field selection during a release, note the dates before and after the change. Good evidence is specific about what it can prove and what remains unknown.

Turn findings into a tested remediation plan

Closing Salesforce security gaps requires a small number of actions that materially reduce exposure. Rank findings by the information accessible, number and type of identities affected, ease of misuse, and ability to detect the event. An unused permission set with no assignments is usually less urgent than a public site that exposes customer contact records. Don’t use the presence of Shield as a blanket reason to lower priority for excess access.

Use a 4-column remediation register that connects each finding to proof and a repeatable retest:

FindingEvidence to captureLikely ownerRetest
Overbroad system permissionUser assignment and effective accessIdentity or Salesforce adminConfirm right removed without blocking work
Public object sharingTest record visible across teamsCRM data ownerRepeat cross-team read test
Sensitive field exposedReport or API test from limited userApplication ownerTest field Read and Edit separately
Missing export alertControlled export with no event responseSecurity operationsGenerate event and verify alert receipt
Incomplete change historyField change absent from approved historyCompliance and CRM ownerConfirm new changes are recorded

Document fixes in a Salesforce security assessment register rather than relying on email threads. Preserve a snapshot of the original configuration where your change process permits it. Link each finding to a ticket and include a rollback decision for changes that could affect daily work. A complete record includes the business owner, affected metadata, approval date, change ticket, test user, expected result, and retest outcome. Separate quick access removals from changes that need process redesign. For instance, removing an unused contractor’s permission set is different from rebuilding sharing rules that support a sales territory model.

Salesforce’s Transaction Security controls can limit some dangerous actions while longer repairs proceed. Salesforce’s Transaction Security framework can evaluate supported real-time events and take actions such as blocking or requiring additional verification, subject to licensing and policy setup. Test the policy carefully so legitimate users aren’t locked out during a busy reporting period. Monitoring can reduce risk while a fix is planned, but it shouldn’t become an excuse to leave unnecessary privileges active.

Keep access decisions current after the assessment

The final review should leave a working owner for each major access rule. Tie permission reviews to hiring, role changes, contractor exits, new integrations, and releases. New managed packages can add rights. New reports can make fields available to a wider group. A sales reorganization can invalidate a sharing rule even when no administrator touches it. Recheck the changes that are most likely to expand access instead of waiting for the next annual audit.

A workable schedule gives sensitive permissions more frequent review than stable low-risk settings. Review privileged users and external integrations regularly, test major sharing changes before deployment, and verify critical monitoring alerts after maintenance. Keep a reference set of approved users and test records so the same negative access tests can be repeated. The aim is to catch drift before it becomes an incident.

The strongest result of a Salesforce security assessment is a set of decisions the organization can defend: each privilege has a purpose, each sharing exception has an owner, and each important security event can be investigated. Salesforce security gaps tend to return when permissions are treated as one-time setup. They stay manageable when access is checked against actual work and the evidence remains usable.

Frequently asked questions

What is a Salesforce security assessment?

It is a review of identity controls, effective permissions, record visibility, connected systems, and security evidence in a Salesforce org. The reviewer tests whether access matches approved business needs and records changes needed to close gaps.

How is a Salesforce security assessment different from Health Check?

Health Check compares selected configuration settings against a baseline. A full review also tests user access, sharing exceptions, integrations, code paths, and monitoring. A strong score can’t prove that confidential records are visible only to approved people.

Which Salesforce permissions should be reviewed first?

Start with Modify All Data, View All Data, user administration, API access, export rights, and access to sensitive objects. Then trace how profile and permission set assignments grant those abilities to actual users.

Can permission set muting remove access from every source?

No. Muting limits a permission inside a specific permission set group. Another permission set or profile can still grant it. Check the user’s effective access before treating muting as a complete restriction.

Do organization-wide defaults control field visibility?

No. Organization-wide defaults establish baseline record access. Field permissions control whether the user can read or edit particular fields. Record sharing determines which records are accessible, while field-level security controls field visibility and editing. A broad sharing override may grant record access, but doesn’t itself grant field access. Check object and field permissions separately.

Why can a user see a record that wasn’t shared directly?

Access can come from a hierarchy, public group, sharing rule, team, territory assignment, or related record. Review the record’s sharing details and the user’s broad permissions to identify the actual path.

Are Salesforce sharing settings enough to secure an org?

No. Salesforce sharing settings affect record visibility, but identity controls, object permissions, field rights, integrations, and code execution also matter. A user with broad administrative rights may bypass ordinary sharing limits.

What should a Salesforce permission audit deliver?

It should provide a role-to-permission matrix, a list of excess grants, the reason each finding matters, and retest evidence. It should also name owners for permissions that remain as approved exceptions.

What does Salesforce Shield include for security reviews?

Shield commonly brings advanced event monitoring, encryption capabilities, and Field Audit Trail, depending on licensing. Each part answers a different question, so coverage and configuration need separate checks.

Does Shield encryption prevent authorized users from reading data?

No. Encryption protects stored data in supported ways, but approved application access can still return readable information. Permission design and data sharing remain necessary even when encrypted fields are enabled.

How do you identify Salesforce security gaps in integrations?

List every integration identity and scope. Check which objects and fields it reads, where data goes, and whether its credentials are still needed. Test whether old connections can access records after retirement.

How often should Salesforce access be audited?

Set review frequency by risk and rate of change. Privileged access and external connections often need closer attention than stable low-risk roles. Recheck sharing and automation after major releases or reorganizations.

Can event monitoring stop a risky data export?

Supported Transaction Security policies can react to selected real-time events, including some export activity. Behavior depends on the event type and licensing. Test the policy with approved sample activity before relying on it.

What evidence should be saved after fixing a gap?

Keep the original test, affected access path, approved change, owner, date, and retest result. That record shows both why the change was needed and whether the unwanted access was actually removed.

What is the first step if a Salesforce org has many security issues?

Identify high-impact exposure first: public access, excess administrative rights, sensitive data exports, and active external connections. Assign owners and test urgent removals before reorganizing the entire permissions model.

By

Leave a Reply

Your email address will not be published. Required fields are marked *