Job search
Prepare a One-Page Brag Document Before You Start a Job Search
Build a concise private fact index with dates, evidence, team ownership and sharing holds, then audit a separate sample-rate claim before using it.
Before updating your resume, prepare a one-page index of work you can explain and support. For each selected contribution, record its date window, your actual responsibility, an evidence reference and whether the fact can be shared externally. The page lets you retrieve facts for different applications while preserving which parts of a team achievement were yours.
This is an editorial adaptation for job-search preparation, not a required employer form or a new ATS resume. The two completed examples below use entirely invented people, work records and numbers. Their local consistency checks do not verify anyone's employment or permission to disclose information. As you browse recent roles on LandOffer, use the responsibilities to select relevant facts rather than invent accomplishments to match a title.

Give the page a specific job
Julia Evans's June 28, 2019 article on brag documents describes keeping a work record for recognition and reflection, including contributions that are easy to forget. Her approach allows longer documents and different update rhythms. The one-page job-search index proposed here is our narrower organizational choice, not her mandated length, copied template or promise of a promotion.
The index should answer a retrieval question: which reliable facts can support the application you are preparing? It can point to a longer authorized work record without reproducing it. You might use one entry for a component you implemented, another for an investigation and another for documentation or collaboration. Keep an entry when it helps you recall supported work, even if it never becomes a resume bullet.
Keep this page distinct from material sent to employers. It may contain reminders about unresolved evidence, restricted facts or an outcome that never occurred. Those reminders are valuable preparation, but they are not ready-made public claims. A private heading does not create permission to store confidential information in a personal account or take internal records off an employer's systems.
Do not spend the first session polishing adjectives. Choose a small set of contributions, identify what you actually know and mark what is missing. If the page grows too long, move supplementary explanation into an authorized working record and keep a reference. Shrinking the font until every detail fits makes the index harder to use and does nothing to improve its accuracy.
Record the action, scope and evidence separately
For each entry, distinguish your action from the project around it. “Implemented the validation checks” and “approved the release” describe different responsibilities. A contribution can be important without conferring architecture ownership, management authority or sole credit for the team's outcome. Preserve your actual held title instead of borrowing the target role's title.
MIT's resume guidance connects activity, project and result, and discusses quantitative scope. Use that downstream writing structure only after you have recovered the facts. Your index can retain the evidence qualifiers that a short resume line might otherwise erase: prototype environment, partial rollout, sample size, pending review or a teammate's decision responsibility.
An evidence reference should explain what it supports. A change record can support implementation; a review record can support acceptance; an operational observation can support use within its stated window. One reference does not automatically prove implementation, acceptance and operational use. Note which claim each reference actually supports. “Merged” does not necessarily mean deployed, and “requested” does not mean adopted.
Use a reference token or an authorized location description when a raw link would expose an internal system. Do not copy tickets, screenshots, source code, customer details or private peer feedback into a personal portfolio. If the evidence must remain inside an organization, record only the permitted reminder in the permitted location. If that permission is unknown, leave the proposed external claim on hold.
Completed one-page index: reliability work with open boundaries
Beatriz Soto is a fictional Senior Software Engineer considering a reliability-focused search. Her authored source notes describe three separate contributions in February through May 2026. The completed index below deliberately retains unfinished work. Every observation is stipulated for teaching; no release system, review tool or employee record was inspected.
| Entry and window | Personal contribution and declared evidence | Result boundary and sharing state |
|---|---|---|
| BS-1: February 1–March 31 | Authored retry-cap checks for a release verification harness; E-BS1 refers to invented change and check records | Checks observed for 24 of 30 scheduled releases; six unobserved. Operations approved releases; another engineer owned harness architecture. External wording on hold |
| BS-2: April 1–30 | Drafted six runbook sections; E-BS2 refers to the invented draft-review log | Four reviewed, two pending. Dalen retained playbook editorial ownership. No training adoption measured; external wording on hold |
| BS-3: May 1–31 | Contributed to three defect investigations; E-BS3 refers to invented investigation notes | One fix merged, two investigations unresolved. Team lead prioritized work; deployment not verified. External wording on hold |
The page's header records Beatriz's held title, preparation purpose and review date of October 23, 2026. Its footer records three decisions: prioritize BS-1 and BS-3 for reliability roles; consider BS-2 for documentation-related responsibilities; do not send the page or its source artifacts while external-sharing authorization remains unconfirmed.
This is a finished fact index, not three polished achievement bullets. It retains enough detail to prevent later inflation. Twenty-four observed checks do not become thirty verified releases. Four reviewed sections do not become six published sections. One merged fix does not become three resolved defects or an assertion that production incidents disappeared.
The saved example includes a separately rendered one-page PDF of this synthetic index. Its references are nonfunctional teaching aliases, not links into an employer's systems. It contains no real proprietary information and establishes no actual disclosure approval. A real candidate should retain the same responsibility and uncertainty distinctions while following the rules governing their own records.
Check numbers before turning them into achievements
For a count, record what was counted and when. For a rate, preserve the numerator, denominator and observation window. For a change, retain the comparable baseline and the later measurement. These are different evidence requirements. A large percentage without its underlying definition is difficult to explain and easy to overstate.
Keep the measured population separate from the entire system. A result from forty selected tickets is not automatically the result for every ticket. Selection method matters: a judgmental sample can help examine particular cases without supporting a general population estimate. Reporting 80.00 percent instead of 80 percent would not make a selected sample representative.
Also distinguish a technical contribution from causal attribution. A team metric can change while several people alter routing, staffing or product behavior. Unless the evidence supports the causal claim, describe your action alongside the bounded observation rather than claiming you personally produced the entire improvement. Do not reconstruct a favorable baseline from memory after deciding which percentage would look strongest.
The O*NET Software Developers description, updated in 2026, includes testing, documentation and collaboration among its tasks. That supports considering a wider range of relevant work than feature launches alone. It does not prove that a particular candidate performed any task or that every employer requires the same mix.
Independent completed audit: an 80 percent claim is rejected
Oren Bell is a different fictional candidate, with different work and numbers. He held the title Software Engineer I. During January 1–28, 2026, the support lead selected forty tickets from a population of 240 for a judgmental review. Thirty-two sampled tickets received a first reply within one day; eight received a later reply. The exercise supplies no comparable prior-period measurement.
Oren drafted routing checks for three request categories and helped review the sample. The support team handled the tickets; the lead selected the sample and accepted routing changes. His proposed brag entry says, “Reduced support delays by 80% across all 240 tickets.” The completed audit rejects that statement on three separate grounds.
First, 32 / 40 = 80% describes the sampled timely-response proportion. It does not measure an 80 percent reduction in delay. Second, the forty selected tickets are not all 240 tickets, and the selection does not establish a representative estimate. Third, Oren did not personally handle every ticket or independently own the team's response process.
| Audit field | Completed disposition |
|---|---|
| Number and window | Preserve 32 timely and eight later replies among forty sampled tickets, January 1–28 |
| Unsupported comparison | Reject the reduction claim because no comparable baseline is supplied |
| Population and ownership | Reject “all 240” and sole credit; retain the lead's selection and support team's work |
| Proposed external wording | “Drafted routing checks for three request categories and reviewed sampled response records with the support lead” |
| Sharing decision | Hold even this narrower wording until real authorization is established; do not export raw artifacts |
The proposed sentence is a completed drafting decision, not an approved public statement. Removing the suspect percentage does not automatically make the remaining project details shareable. The example supplies neither employer permission nor a genuine candidate history. It shows how an evidence audit can end with a useful draft and a clear hold point.
Our original local checker executed seventeen assertions across both cases, then reproduced the same JSON outputs in a fresh directory. It checked the 24-plus-six partition, four-plus-two review status, one-plus-two investigation status, sample arithmetic, ownership distinctions and sharing holds. It did not run a release harness, inspect tickets, verify employment or authenticate permission. The executed calculation and the fictional workplace observations remain separate.

Use the index across materials without losing its qualifiers
Choose the relevant fact before choosing its final format. A resume may need a short implementation statement. An interview preparation note may need the decision, collaborators and unresolved outcome. A portfolio may require an original public demonstration rather than any employer artifact. Reusing a fact does not mean attaching the same private index everywhere.
Maintain a stable entry identifier so a later revision can be traced back to the same record. If a proposal becomes implemented, or a review becomes accepted, update the state and record the new evidence date. Do not overwrite an old observation so completely that you can no longer explain which version supported an earlier application.
Unquantified work still belongs when its scope is supported. A carefully reviewed design question, a reproducible defect report or a clarification that prevented repeated misunderstanding can be worth remembering. State the artifact and your role when the outcome is unknown. Do not convert appreciative feedback into a measured productivity or revenue result.
Before moving a line into an application, check four things together: the action happened, its scope is accurate, ownership is fair and the external wording is authorized. A successful arithmetic check covers only part of that review. If evidence is inaccessible or permission remains unresolved, use a narrower supported statement when allowed, select another contribution or leave the claim out.
Finish the first page, then keep it current
A useful first session can end with three selected facts and an honest unresolved item. Set the next review around a real event such as a completed project or an upcoming application, rather than inventing a universal update schedule. The objective is to recover reliable context when you need it, not to maintain a daily scorecard of personal worth.
Beatriz's page now preserves partial verification, review and investigation states. Oren's independent audit prevents a sample proportion from becoming a population-wide improvement claim. Neither document is ready to send merely because it is concise. The next step is to confirm any missing evidence and sharing boundary before drafting outward-facing material.
Choose one relevant position from recent jobs on LandOffer, then select the index entries that demonstrate its responsibilities without expanding your original role. Keep the evidence qualifiers in the working record and preserve the ones that change the meaning of the final claim. You can then explain the selected contribution from its recorded facts, while leaving unsupported outcomes out of the conversation.


