How to Improve Your CELPIP Writing Score in 2 Weeks

Two weeks is not enough time to overhaul your English. But it is enough time to completely change how you approach CELPIP Writing — and that distinction is what matters here.

CELPIP Writing is scored on four criteria: Content/Task Fulfillment (~40%), Coherence/Cohesion (~20%), Vocabulary (~20%), and Language Conventions/Grammar (~20%). Content carries the most weight. This means an email that clearly completes the task with average vocabulary will outscore a beautifully written email that misses a bullet point.

Two weeks is enough time to dramatically improve your Content and Coherence scores. It is not enough time to overhaul your Vocabulary or Grammar — those take months. Focus on the highest-leverage changes.

Day 1–2: Diagnose exactly where you are losing marks

Write one Task 1 email and one Task 2 survey response under timed conditions. Do not look at rubrics while writing — simulate real test conditions. Then score yourself against the official CELPIP Writing scoring criteria (available at celpip.ca).

For each response, ask three questions:

• 1.Did I address every bullet point in the prompt? (Content)
• 2.Can a stranger read this and immediately understand the purpose, situation, and request? (Coherence)
• 3.Is my word count between 150–200 words? (Both tasks)

Most test-takers fail on #1 (miss a bullet), #3 (too short), or both. Fixing those alone can raise a CLB 7 Writing to CLB 8 or CLB 9.

The Task 1 framework (email)

Every CELPIP Task 1 email — complaint, request, suggestion, apology, or invitation — fits the same four-part structure:

• →Opening (1 sentence): State who you are and why you are writing. "I am writing to bring to your attention a problem with my recent order."
• →Situation (3–4 sentences): Describe what happened with specific details — dates, names, product details. Specificity signals CLB 9 content to graders.
• →Impact or Request (2–3 sentences): What went wrong as a result, and what do you want the recipient to do? Use modal verbs: "I would appreciate," "I would like to request."
• →Closing (1 sentence): "I look forward to your prompt response" signals expectation of action — better than a generic "thank you."

The Task 2 framework (survey / opinion)

Task 2 asks you to respond to a survey question — usually "do you agree or disagree" or "which option do you prefer." The structure is:

• →Position in sentence 1, no exceptions. "I strongly believe that..." or "In my opinion, Option A is clearly preferable because..."
• →Reason 1 + specific example (one paragraph)
• →Reason 2 + specific example (one paragraph)
• →Counter-acknowledgement (one sentence: "While some may argue that...")
• →Restate your conclusion in different words

Week 1 daily schedule (30 min/day)

• Days 1–3One Task 1 email per day, timed (27 min). Check: Did I hit 150+ words? Did I include a specific request? Did I use a formal register throughout?
• Days 4–6One Task 2 response per day, timed (26 min). Check: Is my position in sentence 1? Do I have exactly two reasons? Is my counter-acknowledgement one sentence only?
• Day 7Full Writing session: both tasks back-to-back. Treat it as a mock test.

Week 2 daily schedule (30 min/day)

• Days 8–10Focus on your weaker task (whichever scored lower in Week 1). Three responses in three days.
• Days 11–12Speed practice: write each task in 22 minutes instead of 27/26, leaving 5 minutes to proofread.
• Days 13–14Two full Writing mock sessions. No feedback, no checking — simulate real test conditions exactly.

What the AI Writing Checker catches that you miss

Self-review has a ceiling — you cannot fully see your own errors. After each practice response, run it through the AI Writing Checker. Focus especially on its Content feedback: if it identifies a missing task element (like a missing request in a Task 1 complaint), that is exactly the kind of error graders penalise.