What powers every evaluation
No secrets, no borrowed exam papers — just public frameworks, careful engineering, and honest limits. Here is exactly what your ScoreRoute feedback is built on.
Built on public frameworks
Everything we evaluate against is publicly documented by the institutions themselves.
Official exam structures
The published formats of TEF Canada and TCF Canada: task types, timing, length expectations and module structure.
CEFR descriptors
The Common European Framework of Reference (A1–C2) competency descriptions that underpin both exams.
Canadian Language Benchmarks
The public NCLC/CLB competency framework used by IRCC to interpret language results for Express Entry.
Published scoring information
Officially released band descriptions and public score-conversion tables between exam scales and CLB levels.
Our original work
On top of those public foundations, ScoreRoute adds original engineering. We translated public descriptors into detailed, checkable evaluation rubrics — what does 'adequate vocabulary range' actually look like in a 120-word landlord email? Our rubrics answer that question concretely, level by level.
We then built calibrated AI examiner models around those rubrics: prompt-engineered evaluators that analyse task completion, coherence, vocabulary range, grammatical accuracy and register, and map the combined evidence to a CLB estimate. Their outputs are reviewed against sample submissions and refined regularly.
What we do not use
Being clear about boundaries is part of the method.
- No real or leaked exam questions — ever
- No confidential scoring grids or internal examiner materials
- No affiliation with France Éducation international, the CCI Paris Île-de-France or IRCC
- No claim that our estimates are officially equivalent to exam results
From your submission to a CLB estimate
Linguistic analysis
Your text or response is analysed for accuracy, range, coherence, register and task completion.
Rubric scoring
Each dimension is scored against our CLB-aligned rubrics — the same criteria every time.
CLB mapping
The combined evidence is mapped to an estimated CLB band with the reasoning behind it.
Recommendations
The weakest dimensions become your prioritised, concrete next steps.
Known limitations — stated plainly
AI estimates carry uncertainty. A single evaluation can land a band away from what a certified examiner would award, and official exam conditions add variables no practice tool can simulate. That is why ScoreRoute emphasises trends across many attempts rather than any single number.
We continuously calibrate, but we will never present an estimate as a promise. If a score surprises you, read the accompanying analysis — the specific strengths and weaknesses are usually more valuable than the band itself.