Each Examinizer test has 25 multiple-choice questions and takes about 25 minutes. Questions are drawn from a pool and cover all proficiency levels from A1 to C2 on the Common European Framework of Reference for Languages (CEFR).
How scoring works
Your result depends on which questions you answer correctly and at what difficulty level. The test adapts as you progress: stronger answers unlock harder questions, weaker answers bring easier ones. At the end, the system calculates your CEFR level based on your overall performance pattern, not just the number of correct answers.
What CEFR levels mean
A1 and A2 are basic levels: you can understand simple phrases and introduce yourself. B1 and B2 are independent levels: you can handle most everyday situations and express yourself on familiar topics. C1 and C2 are proficient levels: you can understand complex texts and express yourself fluently and precisely.
Most employers asking for "working proficiency" in a language expect B2. Academic programmes typically require B2 or C1. For a full breakdown of each level, see our CEFR levels guide.
Certificate accuracy
Question difficulty is set editorially at item creation: the author assigns a CEFR level, which is reviewed against the Council of Europe descriptors. Production response data are now analysed on a weekly basis to produce item-level statistics and preliminary difficulty estimates, published in our research environment; these do not automatically change the live question bank. Validated, automated psychometric recalibration is the subject of our current R&D programme — see the full process below. The CEFR level on your certificate reflects your performance on that specific test session. If you feel the result does not match your actual level, you can retake the test — each attempt is independent.
What the certificate does not cover
Examinizer tests assess reading comprehension and applied language knowledge. They do not test speaking or writing. For roles requiring verified oral proficiency, a separate speaking assessment is needed.
How questions are selected
Each test draws 25 questions from a pool of 100 or more per language. The pool covers all six CEFR levels, so the test can assess candidates at any point without asking them to pre-select a level.
Questions cover four areas. Grammar: verb forms, tense usage, sentence structure, articles, and prepositions in context. Vocabulary: word meaning, synonyms, and appropriate word choice in context. Reading: short texts with inference and detail questions. Use of English: collocations, idioms, register, and phrasal verbs.
Early responses influence which questions appear next. Answer several B1 questions correctly and the system serves more B2 questions. Miss several and it returns to A2. The final level reflects your overall performance pattern, not just a count of right answers.
How are questions assigned to CEFR levels?
1. Initial CEFR placement. The question author assigns a preliminary CEFR level based on factors such as vocabulary load, grammatical complexity, text length, and other relevant content characteristics.
2. Descriptor review. The proposed level is reviewed against the relevant Council of Europe CEFR descriptors and can-do statements for that level.
Each question currently goes through this two-step, human-led placement process during question creation.
3. Weekly statistical analysis. Separately from editorial placement, production response data are analysed on a weekly basis to produce item-level statistics and preliminary difficulty estimates. These are published in our internal research environment for review. This analysis does not automatically change the live question bank — it is a monitoring and research layer, not a live recalibration mechanism.
4. Current limitations. Automated, validated psychometric recalibration based on aggregated live test-taker responses is not yet part of the production system. The platform does not currently perform continuous item difficulty recalculation, automatically update the live question bank from response data, or maintain validated, production-grade item-level discrimination or IRT parameters.
5. Future statistical calibration. The next stage of our methodology is validated, automated statistical calibration. Our current R&D programme is developing psychometric methods such as Item Response Theory (IRT) to turn the weekly item-level estimates above into a validated basis for automatically recalibrating item difficulty, and, where evidence warrants it, adjusting the initial CEFR placement. This validated recalibration layer is currently under development and is not presented as an existing production capability.
For a full breakdown of each level, see our CEFR levels guide. To see how this differs from our AI Adaptive Test's per-session scoring, see how the Adaptive AI Test works.
AI Adaptive Test methodology
The AI Adaptive Test works differently from the standard level tests. It starts every candidate at B1 and adjusts every three questions based on responses.
After each block of three, Claude (Anthropic's language model) evaluates the answer pattern, including which levels, which skill areas, and what error types suggest, and decides whether to increase difficulty, decrease it, or hold. After 25 questions, Claude produces a final level with a brief explanation of the reasoning behind it.
This gives a more precise result for candidates between levels or with uneven skill profiles, strong grammar but weak vocabulary, for example. The tradeoff is that the result depends partly on AI inference rather than purely statistical scoring.
Business Language Test methodology
Business Language Tests use a separate question set and a higher pass mark. Standard tests require 60 percent to qualify for a certificate. Business tests require 70 percent.
The question set covers professional vocabulary (financial, HR, legal, commercial), appropriate register in business writing, language used in negotiations and meetings, and business reading texts including reports and memos.
Two questions in each Business test are open-response writing prompts, scored by Claude AI on four criteria: professional tone, content accuracy, grammar and vocabulary, and business appropriateness. Each criterion is worth 25 points. Writing accounts for 30 percent of the total score; multiple-choice questions account for 70 percent.
What the certificate is and is not
Examinizer certificates are not accredited by Cambridge, British Council, IELTS, or any government body. They follow CEFR methodology and provide a verifiable indicator of proficiency.
For contexts where law or regulation specifies a named exam, the UK visa English requirement or the German citizenship test, for example, an Examinizer certificate will not substitute. For job applications, professional profiles, and personal records, it provides a useful and verifiable proficiency indicator.
The QR code on every certificate links to a page showing the holder's name, language, level, score, and date. Employers can check it without creating an account.
Calibration methodology: frequently asked questions
Retesting policy
Tests can be taken free as many times as you like. Certificates for the same language and level are limited to one per 30 days. This prevents repeated attempts until a high-scoring session is certified. Earlier retakes cost €3.
Certifying a different level in the same language has no restriction. You can hold a B1 and a B2 English certificate simultaneously.
Last updated: July 2026