People Management
Build a Team Skills Inventory Without Turning Self-Ratings into Rankings
A fictional four-engineer inventory maps three bounded tasks to self-claims, reviewed evidence, assisted practice, older experience, and unknowns. A completed development decision and follow-up show how to fund learning, preserve an access-blocked opportunity, narrow a contribution claim, and keep experience separate from interest or authorization.
Build a team skills inventory around work the team needs to perform, then record self-reported experience, supporting evidence, review scope, and development interest separately. Keep unknowns visible. Do not add self-ratings into a total, sort people by that total, or treat an interest in learning as proof of current capability. Use the inventory to decide what learning and coverage support the team needs.
This LandOffer.ai guide creates a fictional engineering inventory and uses it to arrange specific development work. It is not a skills examination, performance ranking, promotion ladder, or authorization to perform restricted work. A person’s experience, willingness to help, available capacity, and permission to act are different questions.

Define the work before collecting self-ratings
“Localization: four out of five” could mean writing translations, checking placeholders, operating a translation platform, or designing a release process. Those are different activities. Start with the task, expected conditions, and evidence that would let someone understand the experience being claimed.
The GIZ-hosted Competency Matrix guide, produced by the i-Capital Africa Institute, distinguishes required and available competencies, behavioral demonstrations, and proficiency descriptions. Its broader examples also use scored candidate matrices. This article uses the task-and-evidence distinction; it does not adopt those candidate percentages or turn them into an engineering ranking.
For a small team, begin with a few consequential tasks rather than an exhaustive list of technologies. Ask where the next work needs coverage, where knowledge is concentrated, and where someone wants an opportunity. A task may need a primary contributor and a developing backup; every person does not need every skill.
Write how the inventory will be used and who can see it. Explain whether it informs learning opportunities, temporary work allocation, or a formal organizational process. Do not call it development-only and later reuse it silently for a different purpose. Keep restricted evidence in its approved location and link to an appropriate summary rather than copying customer information into the grid.
Separate self-report, evidence, interest, and unknowns
SFIA’s self-assessment guidance notes that people may overestimate or underestimate their skills and recommends connecting claims to workplace evidence. It distinguishes demonstrated experience from future intentions. Its framework has its own descriptions and responsibilities; the simple labels below are local editorial labels, not SFIA levels or certification.
Ask people to describe something they actually did, their contribution, when it happened, the support they needed, and the conditions that would require help. “Our team shipped it” does not identify the individual’s role. A training completion may show attendance or study, while a practice exercise demonstrates only what happened in that exercise.
Use “unknown” when evidence or opportunity is missing. Mark a missing record as unknown so that a person is not denied an opportunity because they have not yet had one. Also distinguish older experience needing a refresh from no experience at all. A confident claim and a cautious claim can lead to different conversations without making either person more valuable.
Invite development interests in a separate field. Someone may be competent at a task they do not want to become their permanent specialty. Another may want to learn something they have not yet had a chance to attempt. Neither preference should inflate or reduce the experience record.
Worked inventory: four engineers and three bounded tasks
All people, work records, reviews, and results in this example are fictional. Oakframe’s four-person desktop-client team wants to plan learning for three routine tasks. Manager Rina states that the inventory will support the next month’s development assignments. The four engineers can see and correct it; Rina and the named task reviewers maintain it. It is not a production access register or a performance decision.
The task definitions are specific:
- L10: resource review. Compare a proposed interface-string change with its reference, identify missing placeholders, and refer plural-form questions to the localization owner. This does not include approving translations or releasing the application.
- F20: fixture maintenance. Update an existing test fixture for an agreed input change, explain the expected behavior, and obtain the normal code review. This does not establish authority to change the test’s intended behavior.
- R30: release-note traceability. Connect a draft note to its accepted work record, identify unsupported claims, and route unresolved wording to the release owner. This does not authorize publication.
The November 2 inventory uses SC for a self-claim awaiting evidence review, ER for evidence reviewed within a stated boundary, P for practice-only evidence, REF for older experience needing refresh, and U for unknown. These are evidence states, not ordered proficiency scores. A person can have a useful reviewed record for one condition and still need help with another.
| Engineer | L10 resource review | F20 fixture maintenance | R30 release-note traceability | Stated development interest |
|---|---|---|---|---|
| Cora | SC, E1: reports two prior reviews; documents not yet examined | ER, E2: reviewer checked one recent fixture contribution and explanation | U: no record supplied | Wants to deepen fixture-review work; does not offer to own localization |
| Dev | ER, E3: reviewer examined two recent placeholder-review records; plural decisions excluded | U: not discussed in this inventory | SC, E4: reports drafting notes; release-owner review pending | Will help someone practice L10 within agreed time |
| Eli | U: no opportunity to attempt the defined task yet | P, E5: completed a training fixture with assistance | ER, E6: reviewed draft traced to accepted work; publication not included | Wants an L10 learning opportunity |
| Finn | REF, E7: related work from May 2025; current resource format differs | SC, E8: describes past fixture work; supporting record requested | U: no current evidence | Interested in refreshing L10, subject to access and workload |
The evidence index makes the labels interpretable. E1 names Cora’s own claimed contribution and its two record references, marked unread by the reviewer. E2 records that reviewer Jules examined Cora’s October 21 fixture change and explanation on November 2, with the original task scope retained. E2 establishes the contribution Jules examined within its task scope; it does not establish independent maintenance across all fixtures.
E3 records localization owner Anika’s November 2 review of Dev’s October 12 and 26 placeholder checks. Both concerned existing string patterns; neither required a plural-form decision. E4 links Dev’s claimed draft notes but has no release-owner response yet. E5 records assisted training, not shipped code. E6 records release owner Omar’s review of Eli’s October 28 traceability draft, with publication explicitly outside the observation.
E7 retains Finn’s older resource work and its date rather than deleting it or presenting it as current. E8 remains a self-report until the requested evidence is examined. All these records are fictional evidence summaries, not fabricated files represented as actual employer artifacts.
Use the matrix to choose an opportunity, not a winner
Rina initially considers assigning L10 learning to the person with the broadest-looking row. The grid instead shows a narrower operational need: Eli wants an opportunity, Dev has relevant reviewed experience for existing placeholders, and Finn needs a current-format refresh before his older experience can support a coverage decision.
The Polish Integrated Qualifications System’s Skills Audit Method description connects identified competencies and evidence with further development planning. That broad educational and career context supports using evidence to plan learning; it does not validate Oakframe’s labels or make this example a formal qualification assessment.
The completed development decision is:
| Decision | Accepted arrangement and boundary |
|---|---|
| Eli’s L10 practice | Eli accepts 60 minutes across two sessions using fictional, non-production string changes; Dev accepts 45 minutes to explain and review |
| Finn’s refresh | Finn accepts 30 minutes to compare the current resource format with his older example, after approved read access is confirmed |
| Time made available | Rina defers one hour of Eli’s internal cleanup, 45 minutes of Dev’s routine document formatting, and 30 minutes of Finn’s internal index work |
| Review authority | Anika will decide whether further task evidence supports a broader L10 assignment; Dev’s practice feedback alone does not grant release or translation authority |
| Checkpoint | November 16: record what was actually attempted, assistance used, remaining uncertainties, and whether the planned opportunity occurred |
Rina confirms Anika has accepted a separate 15-minute checkpoint review, displacing an internal reference tidy-up. That gives each participant an explicit allocation rather than assuming that someone with expertise is automatically available to coach. The four allocations total 150 person-minutes: Eli 60, Dev 45, Finn 30, and Anika 15.
Cora’s E1 remains unreviewed, and she is not penalized for that. Rina arranges to revisit the outstanding evidence request rather than infer that Dev is “better at localization.” The assignments answer a current learning need and a refresh question. They do not decide salary, seniority, overall contribution, or who should be promoted.

Update the evidence after the opportunity
In the fictional November 16 review, Eli and Dev report that they used the allocated sessions. Eli identified an omitted placeholder in one practice change, initially missed a second mismatch, and corrected it after Dev’s prompt. A plural-form example raised a question they could not resolve; they stopped and referred it to Anika. No actual code, application, or translation system was tested for this article.
The updated L10 cell reads: “P, November 16: two assisted practice examples; one correction after a prompt; plural question referred. Current independent task performance not established.” Interest remains “wants further L10 practice.” Rina does not change the cell to ER merely because the sessions happened or because Eli enjoyed them.
Anika’s checkpoint notes confirm that the examples support planning another supervised task within the same narrow scope. They do not establish repeatable independent performance. Eli accepts that next opportunity in principle; its work item, time, and reviewer still need scheduling, so no production assignment is recorded as accepted.
Finn’s access was not approved in time. His refresh did not occur, and his cell stays REF with “opportunity blocked by access; no new observation.” Rina records IT access owner Casey on the request and asks Casey for a decision date. A blocked prerequisite is not rewritten as lack of motivation or a failed skills assessment.
Cora corrects the description of E1: she reviewed the placeholder section in one change and helped locate a reference in the other. Rina updates the self-claim’s scope without treating the clarification as misconduct. E4’s release-owner review also remains pending. The revised inventory preserves the unresolved evidence and the narrower contribution.
Keep evidence current and its use explainable
Give each record a task definition, source, observation or self-report date, reviewer where applicable, scope, and next review condition. When the task changes, revisit the definition and affected records. Retain an older observation with its date and conditions, and confirm current requirements before using it to plan work in a changed environment.
Do not produce an average of unrelated tasks. A person’s strong release-note contribution cannot numerically cancel missing evidence for a resource review, and counting more skills does not establish greater organizational value. Look for task-specific coverage risks, appropriate learning opportunities, and capacity to support them.
Before sharing a summary more widely, confirm its purpose and audience. Share the minimum evidence needed, preserve corrections, and explain what is still unknown. Ask people whether the record accurately describes their contribution and interest; agreement with the wording does not turn self-report into independently reviewed evidence.
Start with three tasks the team actually needs, define their boundaries, and complete the evidence conversation before planning development work. Keep interests, opportunity gaps, reviewed experience, and authority separate when you update the result. More workplace guides are available from LandOffer.ai.
Sources and Further Reading
- i-Capital Africa Institute / GIZ-hosted Competency Matrix Development and Usage Guide — components on PDF page 8 distinguish required and available competencies and behavioral demonstrations; scored candidate examples are not adopted here.
- SFIA Foundation: Self-assessment guidelines — checked October 27, 2026; workplace evidence, personal contribution, and distinguishing demonstrated experience from future intentions. No SFIA level or certification is assigned in this article.
- Integrated Qualifications System: Skills Audit Method — page updated October 9, 2026; identified competencies, evidence portfolios, and further development in its stated context.



