1. LandOffer
  2. Blog
  3. All articles
  4. Salesforce Portfolio for Job Applications: Show a Business Workflow Beyond Badges

Job search

Salesforce Portfolio for Job Applications: Show a Business Workflow Beyond Badges

A complete fictional quote-review design and an executed local model show business scope, proposed controls and bounded verification beyond learning badges.

10 min readLandOffer team

A Salesforce portfolio for job applications should make one business workflow understandable: who needs it, what you personally designed or configured, which data and permissions it uses, and how you checked whether the result meets the request. Keep learning badges as supporting evidence of study. They do not replace a case that explains your decisions and their limits.

LandOffer publishes this guide and offers job-search tools. As you browse recent roles on LandOffer.ai, compare the work requested with the evidence your portfolio contains. All business records, candidate actions and organizations below are fictional teaching examples. Our executed work is an original offline Python model; no Salesforce organization, account, badge, configuration, customer data or approval was accessed or changed.

A business request branches toward two review trays while learning tokens remain a separate supporting folder.

Give the reviewer a business problem to follow

Choose a small workflow with a visible beginning, decision and handoff. A sales exception request, service follow-up or equipment record can supply that structure when the records are synthetic and the scope is explicit. An attractive dashboard without a defined user decision leaves the reviewer guessing why its fields and calculations exist.

Describe the user’s request and decision before naming Salesforce features. Identify the requester, the person who decides the business rule, the intended operator and the recipient of the output. Record what is excluded. A practice project should not quietly expand into invoicing, customer communication or production deployment because those additions make its screenshots look substantial.

Then describe your contribution at its actual stage. A proposed field model is a design artifact. A configured sandbox is an implementation artifact. A recorded test result concerns the environment and cases tested. A stakeholder’s acceptance concerns an agreed scope. These are useful, different forms of evidence; preserve the distinction wherever the case appears.

For an application, give a reviewer an obvious reading path: request, personal decisions, data and controls, checks, result, remaining work. A short case summary can link to permitted supporting files. Avoid asking a hiring team to navigate an entire training history before discovering what you built.

Show what learning does and does not establish

Salesforce’s February 2024 explanation of Trailhead hands-on learning distinguishes interactive quizzes from hands-on challenges and describes projects that guide practice in a Trailhead Playground. Such learning can help develop familiarity. Its completion does not, by itself, document your independent requirements analysis or a business owner’s acceptance of your own project.

Name only learning you actually completed. Keep a guided exercise labeled as guided, including what came from its instructions and what you added. Do not present a copied challenge solution as an independently commissioned business system. If a particular learning or assessment policy restricts sharing its tasks or answers, respect that boundary and create your own demonstration instead.

A portfolio can show both learning and independent reasoning without devaluing either. Put the relevant learning near the project as background. Spend the case’s main space on a choice you made, an alternative you rejected, a negative test you included and the result you can substantiate. This helps a reviewer distinguish knowing a feature name from deciding when and how to use it.

Worked asset 1: a complete fictional quote-exception case

Nabil, the candidate, and Juniper Parcel Supply, the business, are wholly invented. This is an illustrative design case, not an implemented Salesforce solution, real customer engagement or tenant screenshot. Its business owner’s rules are stipulated teaching facts. No actual owner approved them.

The fictional sales team needs to review proposed quote discounts before a separate quoting process continues. The requested demonstration covers exception intake and reviewer eligibility only. It does not issue a quote, calculate tax, send email, reserve stock or update accounting. Approval eligibility in the model must not be described as a completed commercial approval.

The filled case record is:

Case component Completed example and boundary
Business request Let a representative submit a proposed discount and reason, then identify an eligible independent reviewer. Hold unsupported requests rather than silently sending them onward.
Synthetic data Request DEMO-QEX-A; requester DEMO-REQUESTER; state Submitted; discount 11%; reason “Synthetic seasonal stock request.” These labels are not Salesforce record IDs.
Stipulated decision rule Sales leads may review 0–10% inclusive; pricing managers 0–20% inclusive. A requester cannot review their own request. Drafts and blank reasons remain held.
Proposed configuration A custom request object with state, requester, discount and reason fields; a submission interface; eligibility logic; and a review queue. All remain design proposals, not configured features.
Proposed access Representatives create and read their own requests; designated reviewers read assigned requests. Review-decision fields belong to reviewers; representatives cannot edit them. Object, field and record access need separate tenant verification.
Personal contribution In this fictional case, Nabil authors the business map, field dictionary, proposed access matrix and test specification. In our actual article exercise, AI-generated Python models the stipulated rule offline.
Completed result The local model holds the 11% request for a sales lead and marks an independent pricing manager eligible. No Salesforce request was routed and no business approval occurred.
Remaining acceptance Intended-user tenant tests, effective-access review and real business-owner acceptance have not happened. The case is a bounded design-and-model prototype.

The accompanying proposed access matrix is also filled, rather than reduced to “permissions configured”:

Intended persona Intended record and field boundary Evidence still needed in a tenant
Representative Create a request and read their own; proposed discount and reason editable in Draft. No direct approval-decision edit. Own-record access, unrelated-record denial and decision-field edit denial under the representative’s effective access.
Assigned reviewer Read assigned submitted requests; write the review decision only within the stipulated review policy. Assignment visibility, unassigned-record denial, field access and actual flow execution context.
Unrelated staff member No access to the demonstration request records. Read and update denial through the relevant user-facing paths, not merely an absent navigation tab.

These are intended controls and unperformed checks. The independent-review business rule still needs enforcement even when a person has a reviewer role. The local model below examines that rule; it cannot confirm these object, field or record boundaries.

The completed portfolio decision is to publish, if permitted, a design-and-model case under independent projects, with the implementation gap visible. Its summary can accurately say: “Modeled a synthetic discount-review workflow, documented proposed data and access controls, and executed finite local eligibility checks; Salesforce implementation and business acceptance remain unperformed.” It cannot say “deployed a secure approval system.”

For a real project you implemented, replace those proposed labels only where you have the corresponding evidence. Include the actual environment and your role, redact private identifiers and retain who approved the business rule. Do not copy Nabil’s invented contribution into your own resume or portfolio.

Make permissions and execution context part of the case

Trailhead’s object-access guidance describes profiles, permission sets and permission set groups, and distinguishes object, field and record controls. A proposal that grants access to a request object has not fully explained which requests a user can see or which decision fields they can change.

Use your case to state intended access and show the checks that would establish effective access in the actual environment. A representative seeing their own request is a different test from a reviewer editing a decision. Include an unrelated user and an unauthorized field change as negative cases when appropriate. An administrator screenshot leaves the reviewer unable to assess whether the intended user can perform the task and an unrelated user is denied.

Salesforce’s flow-debugging lesson demonstrates that running-user access and flow context affect behavior. It also shows that debug runs can update records and discusses rollback mode. Those details make “I clicked Debug” an incomplete test description. Record the environment, user context, input, expected result and whether a run was intended to persist changes.

Do not solve an access failure by broadly weakening controls just to obtain a successful screenshot. Explain the intended user task and the business owner’s approved scope first. A local conditional function cannot determine whether a Salesforce flow respects the proposed access matrix, uses an appropriate context or interacts safely with other automation.

The example here makes no tenant test or security claim. Its proposed interface, permissions and review queue would need implementation and independent verification before the project could be described at that stage. This missing work is a concrete portfolio limitation, not something a badge count can close.

An abstract test gate keeps an independent reviewer separate from the requester and marks a rejected model case.

Keep failed cases in the finished evidence

A portfolio test section should show how a failure changed your decision. Include the input and expected business result, then report the observed result at the level you actually tested. Include a rejected case so a reviewer can see which boundary the checks challenged.

Worked asset 2: the executed local rule model and decision ledger

We wrote and executed an original Python model using only synthetic records. It considers three roles, four discount values, two requester/reviewer relationships, two request states and two reason states. That produces 96 combinations: 3 × 4 × 2 × 2 × 2. Six satisfy the stipulated model rule; 90 are held. These are counts of a finite test space, not business approval rates or a Salesforce security assessment.

The executed expected-result check uses a separately enumerated list of allowed role/percentage pairs, plus the submitted, reason-present and non-self-review conditions. This is an independently expressed oracle inside the same AI-authored teaching exercise, not an external business-policy certification. This helps expose a mismatch between the function’s branching and the written test policy. It does not prove that the policy itself is suitable for a real business.

The completed readable ledger includes these cases:

Synthetic input Local observed result Completed interpretation and decision
Independent sales lead; Submitted; reason present; 11% Held above review limit Keep the request unapproved for this reviewer. The lead’s authority ends at 10%, not at the overall 20% ceiling.
Independent pricing manager; same 11% request Eligible for model approval The model permits this reviewer to decide. It does not record that the manager approved a commercial quote.
Requester acting as pricing manager; same request Held for self-review A high-authority role does not remove the independent-review requirement.
Independent pricing manager; Draft; 11% Held as not submitted A valid reviewer cannot make an unsubmitted request ready.
Independent pricing manager; Submitted; blank reason Held for missing reason Retain the missing-information state rather than treating reviewer availability as sufficient.

Additional assertions hold negative and 21% discounts and permit exactly 10% for an independent sales lead with a submitted, reasoned request. A deliberately overbroad alternative treats 11% as eligible for a sales lead because it uses the overall 20% ceiling. The correct model holds that same input above the lead’s 10% limit, so the counterexample rejects the alternative. The original synthetic input remains unchanged after checking.

The completed decision is to retain the narrower lead rule in the local model and record the overbroad version as rejected. The artifact bundle contains the policy, code, synthetic inputs and execution results. It does not contain a Salesforce flow export or screenshot because we did not create one. Passing these cases proves neither effective permissions nor full correctness for arbitrary inputs, concurrency or subsequent edits.

A real implementation would also need to define what happens when an approved request changes, a reviewer loses access or automation is retried. Those behaviors are outside this finite exercise. Name them as unresolved requirements rather than describing the model as production-ready.

Package the case for an application

Lead with the business need and the stage reached. Give reviewers a short index to the field model, proposed access matrix and actual test evidence, with synthetic labels visible. Keep credentials, organization URLs, private client records and access tokens out of public materials. Review whether every file is yours to share before making a case accessible.

Choose the strongest bounded project rather than inflating the number of features. A reviewer should be able to identify your contribution, distinguish a modeled decision from a platform result and see why an unsuccessful case was held. That gives an interviewer specific decisions and evidence limits to discuss alongside your learning record.

Before attaching a portfolio link, check one complete case against your actual contribution, environment, evidence and remaining work. Then compare those responsibilities with the target vacancy. Explore LandOffer.ai to organize the next application while keeping your portfolio claims tied to evidence you can explain.

Sources and Further Reading