Short answer
A skills gap analysis tool should compare a specific work requirement with dated evidence of current capability, then route the result to development, sourcing or process change. Start with the task and its acceptable output. Label self-report, demonstrated work and verified records separately. Keep unknowns visible, and require a human reviewer before acting on any individual conclusion.
Use this guide to run the analysis
This guide is a working sheet, not a software ranking. Use the enterprise talent mapping guide when the job is to map capability, employee interest and development options across a business decision. Use the workforce capacity worksheet when the main question is whether enough qualified hours are available at the right time.
The CIPD learning needs analysis factsheet recommends starting with current and future capability needs, assessing existing skills and knowledge through formal and informal methods, and interpreting gaps within organisational strategy. It also cautions against defaulting to a course before diagnosing the performance need.
Define the work before naming a skill
Broad labels such as communication, digital literacy or leadership are difficult to verify and too vague to guide action. Rewrite the requirement as observable work with conditions and an acceptable output.
| Vague label | Testable work requirement | Conditions to record | Acceptable evidence |
|---|---|---|---|
| Data analysis | Update a six-week demand forecast and explain the changed assumptions | Approved dataset, spreadsheet tool, two-hour review window | Recent accepted forecast or comparable work sample |
| Customer communication | Explain a delayed order and agree the next step | Defined policy, accessible channel, realistic case | Reviewed case note or structured role-play |
| Team leadership | Run a shift handover that identifies owners and unresolved risks | Actual role authority, handover template, shift context | Observed handover and completed record |
Public taxonomies can help teams use consistent terms. The US Department of Labor sponsored O*NET database separates skills, knowledge, tasks, work activities and work context. The European Commission's ESCO occupation profiles link occupations with relevant knowledge, skills and competences. Use either as a vocabulary starting point, then validate the requirement against the work in your organisation. A taxonomy entry does not prove that a person can perform a local task.
Copy this evidence worksheet
Use one row for one work requirement. Split a role into several rows when the tasks need different evidence or actions.
| Field | What to record |
|---|---|
| Business need | Decision, service or change that creates the requirement |
| Work requirement | Task, conditions and acceptable output |
| Needed by | Date and expected duration |
| In-scope group | Roles or teams included, with reason |
| Existing evidence | Source, date, reviewer and what it actually shows |
| Evidence state | Unknown, declared, described, demonstrated or formally verified |
| Limitation | Missing context, old evidence, different tool, access issue or unresolved disagreement |
| Gap type | Capability, capacity, access, process, role clarity or evidence gap |
| Proposed action | Development, sourcing, work redesign or further evidence collection |
| Owner and review | Accountable owner, next check and completion evidence |
Avoid one overall skills score. Two pieces of evidence may answer different questions, and averaging them can hide that difference. Preserve the source and limitation instead.
Distinguish self-report from demonstrated work
Each evidence source has a useful job. None should silently become a verdict about the person.
| Evidence source | What it can support | What it cannot establish by itself | Useful next check |
|---|---|---|---|
| Employee self-report | Experience, confidence, interest, constraints and examples to examine | Consistent performance under the required conditions | Ask for a relevant example or offer a fair work sample |
| Manager account | Observed work, support given and local expectations | A complete or bias-free view across contexts | Compare a dated example with the agreed standard and employee account |
| Existing work artifact | Output produced in a real setting | Which part the person completed or whether the context is still current | Confirm authorship, conditions and review outcome |
| Work sample or simulation | Performance on a defined task under stated conditions | Performance in every future situation | Use comparable instructions, tools, time and accessibility support |
| Qualification or formal record | Completion of a defined programme or verified requirement | Current application to a different local task | Check recency and relevance to the work requirement |
For recruitment, the CIPD's current selection-method guidance says skills assessments should resemble real tasks as closely as possible. The same design principle makes an internal demonstration easier to interpret, but employment context matters. Define the purpose, use proportionate evidence, provide needed adjustments, and do not reuse the result for an unrelated decision.
Run the analysis in seven steps
- Name the decision. State what work must be delivered, by when and why the evidence is needed.
- Define the requirement. Describe the task, conditions and acceptable output with the people who know the work.
- Reuse appropriate records. Check recent work, qualifications and learning records before asking people to repeat evidence.
- Invite employee context. Ask about relevant work, confidence, interests, constraints and the accuracy of existing records. Keep this labelled as self-report.
- Resolve only material uncertainty. For a high-priority gap, review a recent artifact or offer a relevant, accessible demonstration. Do not test every person because the tool makes it possible.
- Classify the gap. Decide whether the obstacle is capability, capacity, access, process, unclear ownership or missing evidence.
- Choose and review an action. Name the owner, expected evidence, support and review date. Preserve disagreement rather than forcing a confident label.
Work through a fictional example
The following example is fictional and demonstrates the worksheet. It is not a claim about a customer or a benchmark for other teams.
A service company needs eight planners to update a six-week demand forecast, explain material assumption changes and revise it after a promotion. The workflow must be ready before the next planning cycle.
The review starts with evidence already available:
- all eight planners completed the same spreadsheet course;
- eight report that they have worked with forecasts, but their examples cover different tasks;
- four have recent forecasts whose assumptions and review outcomes can be checked;
- two have artifacts created under an old process;
- two have no comparable recent artifact.
Course completion confirms participation in learning. The self-reports identify experience to examine. Neither proves the full work requirement. Human reviewers compare the four current artifacts with the agreed criteria and mark the requirement as demonstrated for those four people.
The other four are offered the same fictional dataset, instructions, reference material, tool access and review window. One produces a forecast with traceable assumptions, so the evidence state becomes demonstrated. Two apply the basic formulas but cannot yet explain how the promotion changes demand, creating a targeted development need. The final person cannot access the reference demand file. That result is recorded as an access gap, not a skills failure.
The usable result is:
| Evidence state or gap | Count | Action | Review evidence |
|---|---|---|---|
| Requirement demonstrated | 5 | Confirm availability for the planning cycle | Accepted forecast under the current process |
| Targeted development need | 2 | Guided practice on promotion assumptions, then a second sample | Revised forecast and explanation reviewed |
| Access gap | 1 | Repair permissions, then offer the same sample | Access confirmed before any capability conclusion |
If the planning cycle needs eight capable people before development can finish, the owner can compare temporary sourcing, narrower scope or process redesign. The worksheet exposes that decision. It does not turn five demonstrated examples into a prediction about future performance.
Choose action by cause and timing
| Finding | Development | Sourcing | Process or work redesign |
|---|---|---|---|
| Skill can be developed before the need date | Practice, coaching, assignment or learning with evidence afterward | Usually unnecessary for this gap | Check whether clearer standards would also help |
| Required capability is absent and the deadline is near | Begin development for later demand | Recruit, contract or borrow verified capability | Reduce or resequence the work where feasible |
| People can perform the task but lack access or authority | Do not prescribe training | Do not hire around a permissions problem | Fix tools, data, decision rights or handoffs |
| Requirement is unclear or evidence is missing | Pause a broad programme | Pause a role specification | Clarify the work and collect proportionate evidence |
Actions can run together. An organisation may source short-term capacity while developing internal capability and redesigning a poor handoff. Record each action's owner and timing so that a training completion is not reported as a closed business gap.
Protect the people represented in the worksheet
Skills evidence can affect development, hiring, mobility and performance decisions. Set the purpose and access before collection. Store only what the decision needs, show the evidence date and source, let people challenge factual errors, and define when the record will be reviewed or removed.
The UK Information Commissioner's Office employment-records guidance covers purpose, minimisation, accuracy, retention, security, transparency and access. Organisations should check the rules that apply to their own jurisdictions and use cases.
Where a conversation may help
A focused conversation can collect employee context, work examples and constraints for a defined requirement. Treat the answers as source-linked evidence for human review. Do not infer a complete skills inventory, assign a personal score, predict performance or let a model decide development, mobility or sourcing.
Explore Lontra’s employee conversations and manager briefs, then test one bounded question through one campaign with up to 30 invitations over 60 days, with no card required. Managers receive a brief rather than raw conversations. It is not the system of record, a work-sample platform or an automated decision engine, and no integration should be assumed.
Finish with a reviewable decision
A useful skills gap analysis names the work, shows what each source proves, keeps uncertainty visible and routes the finding to an owned action. The worksheet is complete when a reviewer can explain why the action is development, sourcing, process change or further inquiry, and what evidence will show whether it worked.
