| Test | Questions | Duration | Language | Scoring | Pass mark | Eliminatory | Ranked | Weight |
|---|---|---|---|---|---|---|---|---|
| Raisonnement verbal | 20 | 35 min | — | 1 pt/question | 10/20 | Yes | Yes | 35 % |
| Raisonnement numerique | 10 | 20 min | — | 1 pt/question | 10/20 (combined) | Yes | No | — |
| Raisonnement abstrait | 10 | 10 min | — | 1 pt/question | 10/20 (combined) | Yes | No | — |
The EPSO digital skills test assesses your practical command of everyday professional digital tools and habits: 40 multiple-choice questions in 30 minutes, taken in your language 2.
It covers the 5 domains of the European Digital Competence Framework (DigComp), from information literacy to solving technical problems.
The test is eliminatory: below 20/40, the application stops there, whatever the rest of the competition.
Among the EPSO competition notices currently published, the digital skills test only appears in the generalist administrators competition (AD 5). Candidates for specialist or linguist competitions do not need to prepare it.
The table below shows the exact parameters of the test in each competition concerned, as published in the Official Journal.
| Competition | Questions | Duration | Language | Pass mark |
|---|
The questions below come from our practice bank, in the format of the real test: one question, four options, a single correct answer, with the full explanation.
Answer key — A
The W3C Web App Manifest is a JSON document describing how a web app should behave once installed on a device. It declares name, short_name, icons (multi-sized), start_url, display ("standalone", "fullscreen", "minimal-ui"), theme_color, background_color and orientation. The browser uses it to render the splash screen and install icon.
A. Correct: PWA install metadata (W3C Web App Manifest).
B. Wrong: the Web App Manifest has no link to cookie consent; consent categories are managed by a separate consent management platform, not manifest.json.
C. Wrong: saved passwords and autofill are handled by the browser's built-in credential manager, not the manifest.json file.
D. Wrong: offline data sync is the role of the Background Sync API and service workers, not the manifest.json file.
Answer key — B
UTF-8 (RFC 3629) is a variable-width encoding (1 to 4 bytes) covering the entire Unicode code-point range. ASCII characters are encoded in 1 byte identical to US-ASCII: full backward compatibility. It is the default encoding recommended by W3C/HTML5. Source: RFC 3629, WHATWG HTML Living Standard.
A. UTF-8 is not a fixed 8-bit encoding; it is variable-width (1 to 4 bytes) per RFC 3629, unlike single-byte Latin-1.
B. UTF-8 (RFC 3629) is a variable-width encoding (1 to 4 bytes) covering the entire Unicode code-point range. ASCII characters are encoded in 1 byte identical to US-ASCII: full backward compatibility. It is the default encoding recommended by W3C/HTML5. Source: RFC 3629, WHATWG HTML Living Standard.
C. A character encoding does not encrypt data; UTF-8 has no cryptographic function, unlike TLS which handles HTTPS confidentiality separately.
D. UTF-8 does not compress text; it can be the same size as ASCII or larger than UTF-16 for many non-Latin scripts.
| Connaissances de l'Union europeenne | 30 | 40 min | L2 | 1 pt/question | 15/30 | Yes | Yes | 25 % |
|---|
| Competences numeriques | 40 | 30 min | L2 | 1 pt/question | 20/40 | Yes | Yes | 25 % |
|---|
| EUFTE (epreuve ecrite) | 1 | 40 min | L2 | 1 pt/question | — | Yes | Yes | 15 % |
|---|
| Weight |
|---|
| Administrateurs generalistes (AD 5)EPSO/AD/427/26 | 40 | 30 min | L2 | 20/40 | 25 % |
The competition notice explicitly ties the test to the European Digital Competence Framework (DigComp) and lists five domains. Every question in the test belongs to one of them.
Domains as listed in the official competition notice (full reference at the bottom of the page).
Answer key — B
Article 50 AI Act = labelling and transparency obligations for AI content.
A. False: Article 50 of Regulation (EU) 2024/1689 specifically imposes transparency duties on synthetic/deepfake content, independent of the Annex III high-risk classification.
B. AI Act Article 50: transparency obligations for systems interacting with humans, synthetic content and deepfakes - explicit labelling required.
C. False: Article 50 requires disclosure/labelling, not deletion of the content within any deadline.
D. False: no printing or physical-copy submission obligation exists under Article 50; the requirement is digital disclosure to the audience.
Answer key — B
Pro tone adjusts to the relationship: 'Dear Ms. Smith' > 'Dear Mary' > 'Hi Mary' depending on closeness.
A. Too informal for a first professional contact with someone unknown to the sender.
B. 'Dear Mr/Mrs [Last name]' is the standard professional English greeting.
C. Inappropriately casual slang, unsuitable for any professional correspondence.
D. Inappropriately casual and overly familiar for someone the sender does not know.
Answer key — B
WCAG 2.2 SC 1.4.3 (Contrast Minimum, AA) requires a minimum contrast ratio of 4.5:1 for normal text and 3:1 for "large" text (≥ 18 pt or 14 pt bold). SC 1.4.6 (Contrast Enhanced, AAA) raises these thresholds to 7:1 and 4.5:1. Luminance follows the WCAG formula on sRGB components.
A. 3:1 is the AA threshold for large text, not normal text.
B. Correct: 4.5:1 (normal text AA, WCAG 2.2 §1.4.3).
C. 7:1 is the AAA threshold (1.4.6), not AA.
D. 21:1 is the theoretical maximum (black/white), not a normative threshold.
Answer key — C
Per NIST SP 800-145, SaaS = the consumer uses the provider's applications running on a cloud infrastructure, without managing the underlying infrastructure or platform. Microsoft 365 and Google Workspace are canonical examples. IaaS exposes VMs/network/storage; PaaS exposes an application runtime (Heroku, App Engine).
A. A IaaS delivers raw resources (VM, storage) - the customer manages OS and middleware. Not the case here.
B. B PaaS exposes an application runtime where the customer deploys code - still too low-level.
C. Per NIST SP 800-145, SaaS = the consumer uses the provider's applications running on a cloud infrastructure, without managing the underlying infrastructure or platform. Microsoft 365 and Google Workspace are canonical examples. IaaS exposes VMs/network/storage; PaaS exposes an application runtime (Heroku, App Engine).
D. D on-premise means hosting at the customer site, the opposite of provider cloud.
The costliest mistake is neglecting the test because you "use a computer every day". The questions target precise good practices, licences, data protection, source-reliability criteria, that daily use alone does not formalise.
Second trap: time. At 40 questions in 30 minutes, you have less than a minute per question; scenario questions (a context, then the question) are best read by first spotting what is actually asked.
Finally, the test is taken in your language 2: the digital vocabulary in that language (licence, backup, authentication…) must be acquired before the day, not discovered during it.
Every question in the bank is mapped to one of the 5 official domains: your dashboard shows, domain by domain, where you lose points.
Spaced repetition (the FSRS algorithm) automatically reschedules missed notions at the right moment, instead of making you rework what you already know.
The readiness index (ERI) aggregates coverage, regularity and success rate to tell you whether you are ready, on this module as on the other tests of the competition.
40 multiple-choice questions in 30 minutes, according to the official competition notice (reference at the bottom of the page).
20 points out of 40. The test is eliminatory and its score counts for 25% of the final mark used for ranking.
In your language 2, the one declared when applying to the competition, not in your main language.
No. Among the notices currently published, only the generalist administrators competition (AD 5) includes it. Checked notice by notice, the up-to-date list is in the table on this page.
The European Digital Competence Framework (DigComp), cited by the competition notice, with five domains: information and data, communication and collaboration, content creation, safety, problem solving.
Work domain by domain rather than chaining full tests: identify your weak domains with a diagnostic, then train them separately before testing again under real conditions.
Get upcoming competition notices and calendar updates.
Sign-up opens soon.