Digital skills for work: what they are and how to develop them

1. What digital skills mean

DigComp defines digital competence as the confident, critical and responsible use of digital technologies for learning, work and participation in society. Its 21 competences are organized into five areas: information and data; communication and collaboration; content creation; safety; and problem solving. It is a reference for structuring needs, not an automatic certification of a person or company.

At work, competence depends on context. An assistant may need to register cases, organize files and protect data; a supervisor may interpret a dashboard and approve permissions; a sales team may evaluate information, collaborate in CRM and recognize a fraudulent request. Knowing one feature does not demonstrate that the complete task is performed well.

• Knowledge: understand concepts, risks, sources and rules.
• Skill: perform a task and adapt when the case changes.
• Attitude: act with judgment, responsibility, curiosity and willingness to learn.
• Context: tool, role, data, accessibility, impact and available support.

A competent person does not need to know everything: they can proceed safely, recognize limits and seek useful support.

2. Diagnose tasks rather than label people

Begin with the outcomes the team must deliver. Describe five to ten frequent or critical tasks, the tools used, the data involved and errors that create rework or risk. With consent and practice data, observe a sample: finding a file, checking a source, sharing a document, recovering a version or reporting a problem.

Combine evidence. Self-assessment reveals confidence and perceived barriers but may overestimate or underestimate performance. A practical task shows behaviour in one scenario but not every workplace pressure. Interviews with supervisors and support expose system conditions such as confusing permissions, contradictory instructions, slow devices or processes that force duplicate entry.

• Task, outcome and observable quality criterion.
• Frequency, impact of error and required independence.
• Access to device, connectivity, account, language and support.
• Difference between a skill gap and a system or process barrier.

3. Design training around the actual workflow

OECD work documents how SMEs face time, cost and organizational barriers to training and how much digital learning occurs informally or among peers. This supports short, flexible formats, but it does not mean leaving everything to trial and error. A useful program selects priority tasks, provides deliberate practice and creates a support route after the session.

Use a simple sequence: demonstrate the outcome and criteria, complete a case together, allow individual practice, compare decisions and repeat with a variation. Use test accounts and fictional data when sensitive information is involved. Provide more than one way to access instructions and never ridicule questions or pace; psychological safety also affects willingness to admit a digital error.

• Objectives written as actions participants will be able to perform.
• Routine cases, exceptions and common workplace errors.
• Concise guides available within the workflow.
• Spaced reinforcement and a named person for questions.

A smooth demonstration proves the facilitator can use the tool; practice reveals whether the participant can apply the judgment.

4. Teach information, data and collaboration together

Finding information requires framing the need, selecting terms, identifying the publisher, checking date and evidence, and deciding whether the source fits the task. The result then needs organization: a meaningful name, agreed location, version, owner and review date. Dropping a link into chat without context rarely completes the work.

Digital collaboration adds channel, audience and permission decisions. Teams should distinguish sending a copy from sharing a living source, understand who can view, comment or edit, and close access when responsibilities change. Practise preparing a cross-functional document, recovering a version, resolving duplicates and transferring ownership without losing history.

• Information need and criteria for accepting a source.
• File, folder and version conventions plus an authoritative source.
• Channel selected for urgency, sensitivity and record needs.
• Minimum permission and review of shared access.

5. Integrate content creation, accessibility and safety

Creating digital content includes editing, combining and communicating information within rights and responsibilities. A spreadsheet should allow formulas to be checked; a form should ask only for necessary data; a presentation should preserve contrast and structure; a document should use headings and alternative text where appropriate. Quality is not decoration: it helps another person understand, reuse and review.

Safety belongs inside every task. Practise unique passwords through an approved manager, multifactor authentication, updates, permissions, data handling and verification of unexpected messages. CISA’s Secure Our World campaign emphasizes recognizing phishing, strong passwords, MFA and software updates. Training supports those actions but still needs technical controls and clear reporting paths.

• Understandable, structured and accessible content for its audience.
• Minimum data, approved location and defined retention.
• Verification through a known channel for urgency, payment or account change.
• Early reporting without punishing good-faith disclosure of a mistake.

6. Teach problem solving and responsible AI use

Solving a digital problem is not randomly trying buttons. The person describes the expected and observed result, identifies what changed, preserves the error message, tries a reversible step and records the outcome. When escalating, include device, application, time, affected case and actions already taken. That context reduces repetition and helps support distinguish an individual incident from a broader outage.

For generative AI, add questions before accepting output: which data was sent, what the system may fabricate, how the result will be checked, who owns the decision and whether the use is permitted. OECD analysis notes that most workers do not need advanced AI skills, but do need digital, data and interpretation capabilities. Practise low-risk uses and preserve human review where output affects money, employment, health, rights or safety.

• Expected result, observed evidence and recent change.
• Reversible step, record and criterion to stop or escalate.
• Verification of claims, calculations, sources and generated files.
• Privacy, permission, transparency and accountability for AI use.

A convincing answer is not sufficient evidence. Competence includes verifying before acting.

7. Measure learning and transfer without overclaiming

Before and after training, ask participants to complete equivalent tasks and use a rubric: outcome quality, safe steps, independence, reasonable time and exception handling. Do not measure speed alone because that may reward unsafe shortcuts. Record which support was used and allow repetition; competence includes knowing how to consult a guide or request help in time.

At work, review signals related to the objective: fewer repeated support requests, better file organization, correct permissions, less rework, earlier reporting or higher workflow completion. Interpret carefully because system changes, workload, leadership and process design also affect these measures. Assessment should improve practice and support, not shame people or drive disproportionate employment decisions.

• Baseline and comparable post-training task.
• Observable criteria shared before practice.
• Follow-up agreed with participants, supervisors and support.
• Decision to reinforce learning, redesign the process or improve the tool.