Job search
A Job Application Asks for a Writing Sample: Choose an Excerpt and Add Context
Choose one relevant writing sample, keep an excerpt self-contained, and add accurate context about audience, authorship, editing, and permission.
When a job application asks for a writing sample, choose a piece that demonstrates the writing the role requires, follows the employer’s instructions, and makes your contribution clear. An excerpt can work if it remains understandable on its own. Add a short context note identifying the audience, purpose, origin, and any editing or collaboration the reviewer needs to understand.
This is a selection task for one requested attachment, not a requirement to build a permanent portfolio. Read the actual request before choosing a document. You can explore roles through LandOffer.ai while comparing their writing responsibilities, but the employer’s own instructions govern what to submit. The complete examples below are original fictional teaching samples, not published work or real candidate submissions.

Read the request before judging your strongest piece
Record the requested topic, audience, format, length, attachment count, and any rules about excerpts, editing, collaboration, or writing assistance. Keep that request beside the selected draft so you can check compliance against the actual wording. A request for one sample is not an invitation to send a collection. A requested analytical memo and a customer-facing explanation test different writing decisions, even when both concern software.
Georgetown’s career guidance emphasizes following the application instructions, labeling an excerpt and its origin, and removing sensitive information. It also advises submitting a sample when requested. Those principles fit this task; they do not establish the rules of any particular employer’s application.
University guides offer fallback length suggestions, but a suggestion is not a universal page limit. If the employer specifies a maximum, follow it and determine whether the context note and references count. Do not shrink type or margins to make an oversized selection appear compliant. If an important instruction is unclear, use the designated clarification route rather than silently substituting your preferred format.
Keep this separate from a timed writing exercise or a new assignment with specific assistance rules. An existing sample demonstrates prior or independently prepared writing under its stated conditions. It should not be represented as an unaided assessment response when that is not how it was produced.
Choose writing that is strong and relevant
Columbia’s writing-sample guidance discusses matching the type of writing to the position and prioritizing the quality of the writing. It also explains how to identify a selection from a longer work. For a tech application, relevance means more than mentioning a technology: the piece should show the kind of explanation, reasoning, or audience adaptation the role needs.
A dense design document may show technical judgment but be unsuitable for a role centered on helping nontechnical customers. A polished announcement may show tone but reveal little about analytical reasoning. Choose the piece whose task is closest to the request, then assess whether it contains a clear point, supporting detail, and a conclusion or next action.
When choosing between a weaker relevant piece and stronger unrelated writing, consider whether you can revise a permitted relevant selection. Preserve honesty about its origin and revision. If you create a new sample and the employer permits that, label it as new and unpublished rather than giving it a fictional publication credit.
Give the excerpt a beginning, a decision, and enough context
An excerpt should not depend on missing definitions, figures, or earlier pages to make its central point. Supply essential context in the note or choose a different boundary. Preserve citations needed to evaluate the argument. If an omitted diagram supplies an essential premise, include a permitted version within the requested limits or select a passage that does not depend on it. Mark omissions rather than joining distant passages in a way that changes their meaning.
American University’s guidance recommends prefacing the sample with context and selecting your own authored excerpts from group work. It also says to retain the bibliography when the sample cites sources. Applied to a collaborative technical document, distinguish your prose from another person’s analysis, code, diagrams, and editorial revisions.
Use the context note to orient the reviewer. Let the excerpt show the quality of the writing rather than describing it with unsupported praise. State what the piece was written to accomplish and what you did. If another person substantially edited it, describe that help accurately and check whether the employer accepts such work. Permission to share is a separate question from authorship: writing a document does not automatically make it yours to disclose.
Finished sample one: a technical decision excerpt
All names, products, requirements, and text in this example are invented. The excerpt was newly written for this guide. It is not a real employee document, candidate achievement, or published article, and it must not be claimed as someone else’s independent work.
Fictional request: A product engineering application asks for one concise example explaining a technical decision to an engineering audience. It permits an unpublished sample with a context note. No universal requirement is implied.
Context cover sheet — Selwyn Hart
Title: Exact tags for a local note filter.
Audience and purpose: Engineers deciding how a fictional note-taking prototype should interpret a tag selection.
Origin: New, unpublished demonstration memo for an invented product called Larch Notes. This selection contains the recommendation, rationale, and limitation; it is not a published clip.
Contribution and editing: In this fictional attribution, Selwyn writes the memo and defines the illustrative requirements. No external editor, user study, or production implementation is asserted. This is a fictional cover-sheet example, not a claim about how a real candidate prepared a sample. In a real application, replace this fictional credit with the actual authorship and assistance history.
Permission and sources: No employer files, customer records, or third-party prose are used. The scenario’s rules are stated in the sample rather than attributed to a real product.
Selected excerpt: Exact tags for a local note filter
“For this prototype, selecting a tag should return notes that carry that complete tag. Selecting art should include a note tagged art and exclude a note tagged only cart. A note may have several tags; it qualifies when at least one complete tag matches the selection. The filter does not search note titles or body text.
“I recommend comparing the selected tag with each stored tag rather than searching for it as a substring. This keeps a tag selection distinct from a general text search. It also makes the result explainable: a note appears because it has the chosen tag, not because those letters occur somewhere else.
“The alternative is substring matching. That would help someone who remembers only part of a label, but it would also include cart when the selected label is art. For this feature, that broader behavior conflicts with the supplied requirement. We can consider partial-label discovery in a separate search control rather than combining two meanings in one filter.
“The prototype contract treats labels as exact stored strings. It does not define case folding, whitespace cleanup, or language-specific normalization. We should resolve those rules before adding a claim that Art, art, and a label with surrounding spaces behave identically. A separate test should cover a note with no tags and a note with several tags, including one exact match.
“This recommendation defines a bounded behavior, not a performance result. No user preference or response-time improvement has been measured. The next decision is whether the product owner accepts exact-tag selection as the intended interaction and which normalization rules belong in the contract.”
Why select it: The excerpt recommends exact-tag selection, rejects substring matching for the stated requirement, and leaves normalization unresolved. Those specific choices let a reviewer inspect the reasoning without inventing product evidence. It can stand alone because the key input and expected behavior are inside the passage.
Rejected alternative: A fictional internal incident report with another engineer’s root-cause analysis. Even if its topic looked more impressive, Selwyn could not use that analysis as evidence of his own writing and reasoning without accurate credit and permission. The original demonstration avoids that dependency, provided the actual employer accepts unpublished work.

Finished sample two: explain a product change to customers
This second original fictional sample demonstrates a different audience and purpose. It is not an actual help article, product policy, or customer instruction. The software behavior is defined only for the invented scenario.
Fictional request: A technical support application asks for one example that explains software behavior to a nontechnical reader. It allows a newly prepared sample. The writer chooses an explanation rather than a technical recommendation memo.
Context cover sheet — Maura Finch
Title: Renaming a shared board does not change access.
Audience and purpose: A workspace member who wants to change a board title without confusing a display name with a permission change.
Origin: New, unpublished help-style sample for the fictional product Birch Shelf. The excerpt includes the behavior, an example, and the boundary of the action.
Contribution and editorial credit: In this hypothetical attribution, Maura authors the prose from the scenario contract below. No collaborator’s text, editorial review, customer feedback, or actual publication is claimed. Any real submitted note must disclose the actual contribution and assistance rather than copy this fictional attribution.
Scenario contract: A board has a stable identifier. Renaming changes its title, not its identifier or membership. An offline view can show an older title until it refreshes. These rules are invented, not promises about real collaboration software.
Permission: This sample uses no private workspace, screenshots, customer information, or employer-owned material.
Selected excerpt: Renaming a shared board does not change access
“A board’s title helps people recognize it. Changing that title does not create a new board or change who can open it in Birch Shelf.
“For example, you might rename ‘Workshop ideas’ to ‘Autumn workshop plan’ when the team narrows the project. The board keeps the same underlying identifier and the same members. A link to that board still identifies the same board; renaming is not a way to remove someone’s access.
“Before changing the title, choose a name that helps the people using the board understand its purpose. If several projects have similar names, include enough detail to distinguish this one. Avoid putting information in the title that should not be visible to the board’s existing members.
“After renaming, an online view can show the new title. Someone looking at an offline copy may still see the old name until that view refreshes. An older title in that copy does not, by itself, mean a second board exists or that the rename failed.
“If your goal is to change who can open the board, review its membership controls separately. Renaming does not add or remove members. If your goal is to move the work into a different board, that is also a different task; the title change alone does not move the contents.
“When you explain the change to a teammate, identify the old and new titles and say that it is the same board. This gives them a clear way to recognize the work without implying that their access or the board’s contents changed.”
Why select it: The sample defines the action in ordinary language, uses a concrete example, and separates naming from access. It anticipates an offline-view misunderstanding without claiming that a real customer experienced it.
Rejected alternative: A fictional marketing announcement about “transforming collaboration.” It might sound polished, but it would not show the requested ability to explain a specific behavior. Maura’s chosen passage gives the reviewer actual explanatory prose to assess.
Check the attachment as a reader would encounter it
Read the context note and excerpt together. Verify that title, origin, dates if relevant, authorship, and revision status agree. Remove comments, tracked changes, hidden notes, and sensitive material before sharing through an authorized route. Redaction must leave the passage coherent; if it destroys the explanation, choose a different sample or prepare a permitted original piece.
Follow the requested file type and size. If you export a document, inspect the exported file for readable text, complete pages, working references, and a sensible reading order. Saving a file does not establish that an application accepted it. If you later submit through an authorized route, keep the receipt or confirmation distinct from your local preparation record. Keep the selected version identifiable so you can discuss the same writing later.
The university pages reviewed here were accessed October 22, 2026; the reviewed text did not display publication dates. Their general advice is not an employer-specific rubric or evidence that any sample secured an interview. This article supplies finished prose examples, not exported application attachments or submission results.
To make your own selection, browse recent roles on LandOffer.ai, identify the writing task in one actual request, and choose a permitted piece that demonstrates it. Add truthful context, check your permission and contribution statements, and read the selected version from beginning to end against the actual request.


