Career Growth
Build a Promotion Packet That Maps Your Work to the Next Level
Map engineering contributions to your employer’s next-level criteria, write a bounded nomination summary, and expose the weakest part of the case.
Build a promotion packet by connecting a small number of work examples to your employer's actual next-level criteria. For each example, explain the problem, your decisions, the result, the supporting record, and what the example still cannot prove. A reviewer should be able to follow both your promotion argument and its gaps without reconstructing your project history.
Start with the current role framework and the nomination process, then write. Strong performance in your present role is relevant, but it does not automatically establish readiness for a different scope of responsibility. This guide from LandOffer.ai focuses on an internal promotion decision. It does not assess your eligibility or predict an approval.

Agree on the target before collecting more accomplishments
Ask your manager which role and level you are discussing, which version of the framework applies, and who can nominate a candidate. Establish the deadline, required format, evidence period, reviewers, and access rules. If your company has no written ladder, ask for a written account of the responsibilities and examples that decision-makers will use. Do not borrow another employer's title definitions and treat them as yours.
Public frameworks can help you ask better questions. GitLab's Senior Backend Engineer description includes larger technical problems, team influence, and mentoring. Those are examples from one employer; they are not a universal senior-engineer checklist. Your employer might distinguish levels through different combinations of autonomy, scope, execution, and organizational influence.
Clarify the difference between preparation and submission. A working packet can expose missing evidence months before a nomination is possible. An official submission must follow the employer's process. GitLab's promotions page currently warns that it is under review and directs readers elsewhere for current information. That is a useful reminder to verify the controlling internal policy rather than copy a public timetable.
Write the target at the top of your working document. For example: “Software Engineer II to Senior Software Engineer, framework dated September 1, review with manager October 30.” Add “nomination not yet agreed” when that is the current state. A precise target prevents a document intended for one level from quietly becoming an argument for another.
Separate a promotion argument from an activity archive
An archive asks what you did. A promotion argument asks which responsibilities your work demonstrates. Begin with two or three substantial examples, then add smaller examples only when they establish a different criterion or repeated behavior. Twenty similar tickets can bury the decision a reviewer needs to make.
Will Larson's promotion-packet guide recommends developing the document with manager and peer input, including project impact, complexity, organizational contributions, and gaps. Its focus is Staff-plus progression. The transferable idea here is to use a working packet to test the case; its proposed contents should not replace your employer's required form.
For each candidate example, write the criterion first. Under it, describe a choice that was yours, why the situation required judgment, what happened, and where the reviewer can inspect support. Preserve the distinction between initiating a proposal, implementing a solution, approving a change, and owning the ongoing service. These responsibilities can belong to different people.
A self-evaluation may cover an entire review period, including routine delivery and growth. A promotion packet selects evidence for a particular level change. A job-search brag document helps you retrieve facts for future external applications. The same record can inform all three documents, but each document makes a different argument and has a different audience.
Worked evidence matrix: a senior-engineer case with a real gap
The following is a fully fictional teaching case. Nia Patel, her employer, records, dates, and observations are invented. The criteria below belong to the invented company, not GitLab or a real promotion committee. No employee records were accessed or promotion outcome observed.
Nia is an Engineer II seeking Senior Engineer. Her team's hypothetical framework expects independent ownership of a moderately sized project, decisions across a team dependency, and support that increases another engineer's autonomy. Her first draft says, “Led an important build-cache migration and mentored the team.” The completed matrix is more selective.
| Target criterion | Completed evidence and personal action | Boundary or gap |
|---|---|---|
| Independent project ownership | April–June: Nia wrote the build-cache migration proposal, compared three cache-invalidation options, and coordinated the staged rollout. Fictional records N1–N3 cover design, decisions, and release acceptance. | Platform engineers approved infrastructure changes; Nia did not own the entire build platform. |
| Decisions across a dependency | May: Nia and the shared-build owner agreed on a cache-key contract after documenting incompatible assumptions. N2 records Nia's proposal and the owner's approval. | One cross-team dependency supports this project example; it does not prove organization-wide influence. |
| Increasing another engineer's autonomy | July–August: Nia paired with Luis on stale-artifact diagnosis, then reviewed his independently written diagnostic exercise. N4 records his successful second exercise. | This is one engineer and one task. The framework's expectation of repeated mentoring remains insufficiently evidenced. |
The first two rows have useful evidence. The third row has an example but also a gap. Mark it “discuss whether this is enough”; keep “one engineer” rather than expanding it to “the team.” Nia should ask whether a second independent example is needed, whether an existing contribution was overlooked, or whether the current evidence already meets the company's interpretation.
The matrix also avoids counting one migration three times as three independent achievements. A single project can demonstrate several responsibilities, but repetition across rows does not create a longer history. Making that dependency visible helps your manager distinguish broad coverage of criteria from evidence that you demonstrated the behavior repeatedly.
For your own matrix, use accessible internal references with descriptive labels. Do not paste customer data or restricted peer feedback into a document with a broader audience. Check permissions from the intended reviewer's access context through an authorized method, or have the document owner confirm them. A link that works for you may fail for the committee.
Turn the matrix into a short, reviewable narrative
A summary should tell the reader what to evaluate before asking them to open supporting records. Name the requested change, the scope demonstrated, the strongest examples, and the unresolved concern. Keep a longer evidence appendix behind it if the required format permits one. There is no universal page count.
GitLab's Promotion Document Style Guide favors specific contributions and meaningful quantities over flattering adjectives. It also recommends explaining the work rather than leaving numbered links to carry the explanation. Apply that discipline to your own facts without copying its sample metrics.
Here is Nia's completed fictional summary. It uses the same evidence as the matrix, not a second invented success story:
I request consideration for Senior Engineer under our September framework. During April–June, I owned the build-cache migration proposal and rollout coordination for the developer-tools team. I compared three cache-invalidation options and recommended a content-addressed cache key with invalidation behavior agreed with the shared-build owner. Platform engineers approved the infrastructure changes; I coordinated the application changes and release checks.
In the six-week observation window after rollout, the stipulated project log records 18 cache-invalidation alerts, compared with 31 in the preceding six weeks. Build volume and alert configuration were not controlled, so I do not attribute that entire difference to my implementation. Design record N1 and release record N3 explain my decisions and the accepted scope.
The dependency agreement illustrates a decision beyond my own component. The stale-artifact diagnostic exercise with Luis shows one instance of helping another engineer work independently. Repeated mentoring is the weakest part of this case. I request feedback on whether that evidence meets the framework and, if not, which additional responsibility would demonstrate the missing behavior.
The summary makes an explicit request without awarding Nia a title she does not hold. It separates a team observation from personal causation. It also gives a reviewer an identifiable concern to address. Removing the final paragraph would make the document more flattering and less useful.

Audit the argument before asking for endorsement
Use a skeptical reading to test each important claim. For Nia, “reduced incidents by 42%” fails twice: the supplied records count cache-invalidation alerts, not incidents, and the comparison does not isolate her causal effect. The arithmetic difference is 13 alerts, but a correct subtraction would not repair the wrong measure or the unsupported attribution.
Her replacement says what the fictional log records, supplies the two windows, and names the uncontrolled factors. These are stipulated observations in the teaching case, not verified production results. The packet can still explain her technical judgment through the design alternatives and dependency agreement. An impressive percentage is not required to make the responsibility visible.
Next test scope. “Company-wide technical leadership” fails because the supplied evidence concerns one team and one dependency. “Coordinated a developer-tools-team migration with the shared-build owner” survives. “Built a mentoring program” fails because one diagnostic exercise is not a program. “Helped Luis complete a stale-artifact diagnostic exercise independently” survives, with a sufficiency question attached.
Finally, test the time dimension. One successful event might support a criterion, while your employer may require a sustained pattern. Record what your evidence spans and ask how that requirement is interpreted. Do not turn six weeks into a year by describing the work as ongoing. If a system remains in use, distinguish continuing use from continuing personal responsibility.
Ask peers for factual corrections, then review with your manager
Give a peer a narrow request: “Does this paragraph accurately describe my role in the dependency agreement? What did you own that I should credit?” That produces a more useful correction than asking for a generic endorsement. Seek input from people close enough to observe the work; do not manufacture quotations or imply approval from silence.
Discuss conflicting accounts directly and preserve unresolved differences. If a collaborator says the architecture decision was theirs, revise the attribution or identify the record that clarifies the disagreement. A packet should not require colleagues to surrender their own contributions for yours to be visible. Use your employer's channels for concerns about process or fairness.
With your manager, separate three questions: Is the evidence accurate? Does it demonstrate the target criteria? Is nomination possible under the current process? A favorable answer to one does not settle the others. Record the answer to each. For an unresolved criterion, name the missing observation, who can assess it, and the next review date; for a process question, name who will confirm the nomination route.
It is frustrating to hear “show more leadership” after months of extra work. Ask what that means before accepting another assignment: which criterion can it demonstrate, what support is available, and how will success be evaluated? If the concern is timing or an organizational constraint, more unrelated work may not resolve it. Record that distinction so your development plan addresses the actual issue.
Prepare the version that will actually be reviewed
Translate the working packet into the current official format. Keep a dated copy, verify the requested role, remove stale claims, and confirm that references resolve for authorized reviewers. Retain qualifications when cutting length. A shorter account that changes “proposed” into “delivered” is an inaccurate account.
Your next action is to take one completed criterion-to-evidence row to your manager and ask whether it supports the target level. Label it a working example if nomination has not been agreed; you are asking for calibration, not implying that a submission is underway. Expand after that discussion clarifies the standard. This guide is from LandOffer.ai; the promotion decision remains with your employer's process, and a well-written packet cannot guarantee it.
Sources and Further Reading
- Will Larson: Promotion packets — practitioner guidance focused on Staff-plus progression.
- GitLab: Senior Backend Engineer — a single employer's published role description.
- GitLab: Promotion Document Style Guide — specific evidence and clear contribution writing.
- GitLab: Promotions and Transfers — carries an under-review notice; verify current internal policy.

