1. LandOffer
  2. Blog
  3. All articles
  4. Content Design Portfolio: Show Language Decisions, Constraints, and Product Outcomes

Job search

Content Design Portfolio: Show Language Decisions, Constraints, and Product Outcomes

Build a content design case that makes language choices, information structure, constraints, personal contribution, and the limits of product evidence clear.

10 min readLandOffer team

A content design portfolio should show how your language decisions helped people complete a product task. Present the problem, the evidence available, the constraints, the alternatives you considered, and your contribution. Include the resulting content in context, then state what you actually know about its effect. Attractive screens alone cannot explain why a label, navigation structure, or error message changed.

Start with the kind of content work the role requires. A terminology project, a service journey, and a set of recovery messages demonstrate different decisions. LandOffer.ai can help you explore roles while you identify which responsibilities your portfolio needs to explain. Follow each employer’s actual portfolio instructions; there is no single required case-study format established by this guide.

A content designer arranges letterpress pieces into a branching task route, with rejected choices and editing tools beside the map.

Build the case around a task and a decision

A useful case begins with what someone needed to do. “Refresh the product voice” describes an assignment, but it does not explain where the language interfered with a task. Identify the point of uncertainty: choosing the right destination, understanding a status, correcting an entry, or deciding what happens after a button press. Give the reader the original content at that point so they can inspect the problem you describe.

Separate evidence from interpretation. A support theme, a research observation, a search query, and a stakeholder preference are different inputs. Record where each came from and what it can establish. An executive’s dislike of a phrase can be a project constraint without being evidence that users misunderstood it.

The GOV.UK Service Manual’s user-needs guidance distinguishes research evidence from assumptions and connects user needs to the work created to meet them. Published in 2016 and last updated in March 2017, it is service-design guidance, not a current hiring rubric. Its useful lesson for a portfolio is traceability: show how an input led to a content decision, and preserve uncertainty where the input was incomplete.

State your role early. Did you audit terminology, write variants, facilitate agreement, design the information structure, or review implementation? Identify who owned research, interaction design, engineering, product approval, and translation. Describe your contribution with specific verbs: audited, proposed, wrote, tested, or revised. Use “we” for a shared decision and explain your part in reaching it.

Show content structure as well as sentences

Content design can change where information lives, what a category includes, and how a term behaves across a journey. A before-and-after sentence may miss that work. Include a compact map of the old and new structure when navigation, page purpose, or terminology was part of the problem.

Content Design London’s account of its website redesign, dated November 21, 2022, describes connecting user stories and acceptance criteria to pages, developing content types, and working through content with collaborators. It also describes plans for continued measurement. This is a first-party process account; it does not supply a universal portfolio recipe or a quantified result for your project.

For your case, choose one decision worth inspecting. Show the original wording or structure, a viable alternative, the chosen direction, and the reason an option was rejected. Keep enough surrounding context to reveal meaning: a button label depends on the action it triggers, and a status depends on the state it represents. Pair each excerpt with the task and relevant state, rather than asking the reviewer to infer them from a cropped phrase. Explain constraints such as an existing component, translation capacity, terminology shared with another product, or a release boundary.

Worked case: clarify borrowing and returning equipment

Everything in this case is fictional: the organization, designer, product, constraints, copy, and decisions. No user study or product launch took place. It is a completed illustrative case narrative, not evidence of a real designer’s results.

Project: Cedar Loom Workshop’s equipment-loan interface.
Content designer: Linnet Park.
Task: A member needs to find a borrowed tool’s return date and understand whether the loan is still open.
Problem hypothesis: The term “checkout” may be confused with a purchase, while “closed” does not explain whether a tool was returned. This is an assumption to investigate, not a reported research finding.

Available inputs and limits: The fictional brief defines three system states: a tool can be reserved, collected, or returned. A prototype inventory shows “Checkout” in navigation and “Closed” on completed loans. No analytics, support records, or participant evidence is available. Linnet can inspect the content-state mismatch, but cannot claim that members actually made mistakes because of it.

Constraints and ownership: Engineering owns the state transitions; the workshop team owns loan rules. Linnet can propose navigation, labels, and explanatory copy, but cannot change return deadlines or invent a renewal option. A product designer owns layout, and no translation review has occurred.

Before: Navigation says “Checkout.” A collected tool appears under “Active,” with a date labeled “End.” A returned tool appears under “Closed.” The page does not distinguish a reserved tool from one already collected.

Chosen structure and copy:

  • Navigation: “My tool loans.” This identifies the content destination and avoids introducing a purchase meaning.
  • Reserved state: “Ready to collect.” Supporting copy: “Collect the tool from the workshop desk.” The interface shows this only when the fictional system reports a confirmed reservation ready for collection.
  • Collected state: “Borrowed.” Date label: “Return by.” Supporting copy: “Return the tool to the workshop desk by the date shown.” The actual date comes from the system, not from copy written into the page.
  • Returned state: “Returned.” Supporting copy: “The workshop has recorded this tool as returned.” This depends on a recorded return event, not merely the member pressing a button.
  • Unavailable date: “Return date unavailable.” Supporting copy: “Check your loan details with the workshop desk.” The design must not substitute a guessed date or imply that the deadline has disappeared.

Rejected option: Rename every state “Open” or “Complete.” It is shorter, but it removes the distinction between waiting to collect and currently borrowing. Linnet keeps the state-specific terms and records the dependency on engineering’s state mapping.

Decision record: “I proposed the loan terminology and wrote the state-specific content. The product designer retained responsibility for layout. The team still needs to confirm that each label maps to the correct system event and investigate whether members understand the terms. No launch or behavioral result is available. Any later research or implementation evidence would need its own source, date, and project status.”

Portfolio outcome statement: “Produced a proposed terminology set and content-state map for reserved, collected, and returned tools, including an unavailable-date condition.” This reports the design output. It does not claim fewer overdue loans, reduced support demand, or improved task completion.

A real case could add authorized research observations and implementation evidence. It would still need to distinguish a demonstrated comprehension issue from the designer’s initial hypothesis, and a released change from an unimplemented proposal.

A content editor compares a status message with its underlying process, following separate paths for completed work and a recoverable input error.

Worked case: explain report exports and recoverable errors

This second fictional case concerns a reporting product, not a navigation rename. Corwin Ellis is the content designer for Bramble Finch Reports. An analyst wants to create a CSV report, understand whether processing has finished, and retrieve the file. The case contains no real customer data, study, experiment, or software execution.

Starting content: A button says “Get insights.” After selection, the page says “Success!” even while the report is waiting to run. A failed request displays “Invalid.” The fictional engineering specification distinguishes waiting, processing, completed, and failed requests, with a separate validation rule for a missing date range.

Scope: Corwin owns the action label, status text, and recovery wording. Engineering owns job processing and retention behavior. The analyst’s underlying data and the report’s business accuracy are outside this content exercise. The team has not established a processing-time estimate, so the copy cannot promise a delivery time.

Completed content specification:

  • Action: “Create CSV report.” Adjacent explanation: “Choose a date range, then create a report you can download when it is ready.” This names the output and separates creation from retrieval.
  • Missing date range: “Choose a start date and an end date.” Show the message with the date-range controls; retain any entered value that the implementation can safely preserve. Do not describe a processing failure as a missing input.
  • Waiting: “Report queued.” Supporting text: “Your request is waiting to start.” Do not use “ready” or show a download link yet.
  • Processing: “Creating report.” Supporting text: “You can return to this page to check its status.” This wording is conditional on the fictional requirement that the request remains visible when the analyst returns; it is not a claim about every reporting product.
  • Completed: “Report ready.” Action: “Download CSV.” The link appears only when the file exists and is available. Retention details remain an unresolved dependency rather than an invented number of days.
  • Processing failure: “We could not create this report.” Supporting text: “Your report is not available. Try creating it again.” This retry instruction requires engineering confirmation that starting another request is supported; otherwise the team needs a different recovery route.

Why this direction: Corwin replaces the vague action with the actual file operation and removes a success claim from an unfinished state. He rejects “Your insights are on the way” because it adds an implied delivery promise without explaining whether a file is ready. The error specification separates information the analyst can correct from a service failure.

The GOV.UK Design System’s error-message guidance likewise distinguishes validation errors from problems the user cannot fix. It recommends explaining the problem and correction, matching field language, and keeping entered answers. These are useful design principles; this fictional specification is not a tested GOV.UK implementation or an accessibility certification.

Experiment record: “Hypothesis: state-specific labels will help analysts distinguish a queued report from a downloadable file. Proposed comparison: the original prototype and the revised prototype, using the same synthetic report task. Review whether participants can distinguish a queued request from a ready file and identify when the download action is available; record confusion and any failures introduced by the wording. Research method, recruitment, and success criteria require agreement before testing.” This is a plan, not a completed experiment or an A/B result.

Portfolio handoff: “I wrote the action and state language, documented dependencies, and separated validation from processing errors. No participants were tested, no production release occurred, and no task-performance change was measured. Next decisions concern retry behavior, retention wording, and research design.”

Report product outcomes at the level your evidence supports

For real work, distinguish the artifact delivered, the change implemented, and the effect observed. A terminology standard approved by a team is an output. Correct labels appearing in the released product are implementation evidence. A change in task completion is an outcome requiring a defined measure and supporting observations.

If you have quantitative results, explain the population, time period, task, and comparison. Identify other changes released alongside the content. A movement in a product metric does not automatically isolate the effect of your words. If the evidence is qualitative, describe the observation and its limits rather than translating it into a percentage.

An unfinished project can still support a case about a difficult decision. State its status, explain what you contributed, and say which dependencies remain. Avoid filling an evidence gap with a plausible success story. A clear account lets a reviewer discuss your judgment, the remaining risks, and what evidence would change the decision. It does not establish how a particular hiring team will evaluate the case.

Make the portfolio itself easy to review

Lead each case with the task, your role, project status, and central decision. Use readable excerpts with captions that explain what changed. Include enough context for an asynchronous reader without reproducing every workshop, content inventory, or screen. A visual UX portfolio may emphasize layout and interaction; a content design case needs the language and structure legible enough to inspect.

Use only work you have permission to share. Redaction must remove private customer information and confidential product details, not just names. If you reconstruct an example or create a speculative exercise, label it clearly and preserve the distinction from shipped work. A password does not itself grant permission to disclose material.

Before tailoring a case, browse recent roles on LandOffer.ai and select one set of responsibilities to compare with your evidence. Choose the case that best explains that work. Before sharing it, check the employer’s actual submission instructions, confirm your permissions, and make the project’s status easy to find.

Sources and Further Reading