12 CELPIP Writing mistakes that cap you at 7

Your CELPIP Writing has been a 7 for two attempts. Not a 5 — a 7. The grammar is mostly clean, the structure is there, and the score doesn't move, which is the most frustrating place to be. The reality from our grading desk: a 7 is almost never "bad English." It's 2–3 specific habits, each harmless-looking, compounding across both tasks. This post is the 12, grouped by the criteria raters actually score, each with an example, the cost, and the fix. Find your 2–3.

If you want a trained grader to find them for you, take a free mock test first to see where you stand → celpipuni.com/free-mock-test.

TL;DR

Most CELPIP Writing 7s come from 2–3 compounding habits, not weak English: burying the request, sitting on the fence, undeveloped reasons, thesaurus vocabulary, connector-stuffing, no paragraphing. Each costs a little on a different criterion — content, vocabulary, readability, task fulfillment — and together they cap you at 7 while your English is 9. Find yours, fix one at a time.

The core idea: mistakes are criterion-specific

CELPIP raters score four dimensions on every response: content/coherence, vocabulary, readability, and task fulfillment. Every mistake taxes a specific line of that rubric. "Bad writing" is not a diagnosis. "Buried request" is — it hits task fulfillment, and there's a known fix.

So the 12 below are grouped by the criterion they attack. Read them as a menu, not a lecture. When one stings, that's your band-capper. Trying to fix twelve things at once is how people stay at 7 for another test cycle — every sentence becomes a minefield of self-monitoring, and fluency dies. A 7 means you're already doing most things right: your fixes are surgical, not general. "Improve your writing" is what kept you at 7. "Kill the connector habit" is what gets you out.

How to find your 2–3

One practice session, honest eyes:

Write both tasks, timed, typed. Task 1 in ~27 minutes, Task 2 in ~26 — handwriting hides your real habits, and typing changes your sentence rhythm. (Timings get revised periodically — confirm current ones on the official CELPIP site.)
Read against this list, one pass per criterion group. Content first (1–3), then vocabulary (4–6), readability (7–9), task fulfillment (10–12). Single-criterion passes catch what general re-reads miss.
Mark the sting. When you recognize your own sentence in an example, mark it. You're not looking for all 12 — you're looking for the 2–3 you actually commit.

And here's the reference table for what each group attacks:

Criterion	Mistakes	What the rater sees
Content/coherence	1–3	Claims without development; buried points
Vocabulary	4–6	Show-words over owned-words; connector chains
Readability	7–9	Choppy or bloated sentences; wall-of-text
Task fulfillment	10–12	Wrong register; survey structure missing

The mistakes, by criterion

Content/coherence

1. Burying the request. Example: "I am writing to bring to your attention a matter regarding the premises which has been of concern…" — request appears in paragraph 3. Cost: task fulfillment and coherence both dip; the rater's first question ("what does this person want?") goes unanswered for 60 words. Fix: the request goes in by sentence 3, always. In practice runs, underline your request sentence — if it's not on the first page-fold, rewrite.

2. Sitting on the fence (Task 2). Example: "Both options have merit and perhaps a balanced approach would serve everyone." Cost: the survey asked you to choose; not choosing caps task fulfillment regardless of language quality. Fix: state your choice in the first two sentences, concede once, restate at the end. The full formula.

3. First-idea development. Example: "Firstly, parks are healthy. Secondly, parks are green. Thirdly, families enjoy parks." Three claims, zero explanations. Cost: content caps around 6–7 — development density (reason + explanation + example) is the criterion. Fix: two reasons, each with a "because" and a "for example."

Vocabulary

4. Thesaurus words you half-own. Example: writing "I commenced to peruse the documentation" when you'd say "I started reading the papers." Cost: an almost-right big word reads worse than a right small one — vocabulary scores precision, not size. Fix: if you've never used the word in a real email, don't audition it at a timed test.

5. Stringing formal connectors. Example: "Moreover, furthermore, in addition to this, additionally, it is also worth mentioning that…" Cost: connectors doing the work of logic. Coherence through meaning ("The main reason… It's also…") scores; scaffolding doesn't. Fix: ban "moreover" and "furthermore" for two weeks. Use "also," "but," "the other thing is." Nothing breaks.

6. Avoiding phrasal verbs. Example: writing "we should investigate the matter" when the natural line is "we should look into it." Cost: CELPIP Writing is everyday English; avoiding phrasal verbs shrinks your range and stiffens register at the same time. Fix: swap one formal verb per practice response for its phrasal twin — look into, sort out, put off, run by, follow up on. That's what a 9 sounds like.

Readability

7. One-sentence-per-thought chop. Example: "The heater is broken. It stopped on Monday. I called the office. No one answered. It is cold now." Cost: five short sentences in a row read as flat and list-like — the band-7 flatness, even with zero errors. Fix: combine pairs — "The heater stopped on Monday, and after two unanswered calls, the unit's been sitting at 14 degrees." Follow a long one with a short one. Rhythm is scored, through readability.

8. No paragraphing. Example: a 180-word Task 2 response as a single block. Cost: the rater can't see your structure, and unseen structure might as well not exist. Fix: Task 1 = 2–3 paragraphs; Task 2 = 4 (choice, reason, reason/concession, restate). Hit return. It's free.

9. Sentences over ~25 words. Example: "Although I understand that the company policy which was explained to me at the time of the original booking may not allow for the refund which I am requesting, I believe that an exception could possibly be made." Cost: one 40-word sentence contains three chances to lose the rater. Fix: the second "and" is your split point. If a sentence needs a breath, it needs a period.

Task fulfillment

10. Off-register. Example: "Dear Sir or Madam, I am writing to formally express my grievance regarding…" to a building manager you've met twice. Or "hey so the heater's busted, fix pls" to a city office. Cost: register is part of task fulfillment — the test is "can you write appropriately to this reader?" Fix: match the reader. Semi-formal is the default: contractions yes, slang no, "Dear Sir or Madam" never.

11. Missing survey structure (Task 2). Example: a Task 2 response that gives opinions on both options with no stated choice, no reasons anchored to a choice, no close. Cost: it reads as a blog comment, not a survey response — task fulfillment caps at 6. Fix: choice → 2–3 developed reasons → one concession → restate. Every time.

12. Answering around the bullet points. Example: the prompt asks you to cover what you've already tried, and your email says "I have already attempted to address this issue" — with no attempt specified. Cost: bullet points are the task. Missing one is missing the task. Fix: during planning, list the prompt's bullet points and mark where each lives in your skeleton.

What good looks like (one mistake, fixed)

The same email moment, before and after fixing mistakes 1, 4, and 10 together — the compounding effect in miniature:

Capped at 7: "Dear Sir, I am writing to formally lodge a complaint regarding the heating apparatus in my residential unit, which has ceased functioning. I would be most grateful if you could look into this matter at your earliest convenience. Thank you for your kind cooperation in this regard."

Moves like a 9: "Hi Dana — the heater in my unit (214) died on Monday, and it's been sitting at 14 degrees since. I called the office twice — no answer either time. Could you get a technician out this week? Any weekday after 3 works. If it can't be fixed by Friday, I'd like to discuss a rent adjustment for the days without heat. Thanks — Reza, unit 214."

Same situation, same writer, same English. The second names the reader, makes the request in sentence 3, uses a number, and frames the consequence as a next step. Nothing fancy — just deliberate.

Your fix-it checklist for this week

Write both tasks once, timed, typed — your diagnosis run
Four passes against this list, one per criterion group
Mark your 2–3; ignore the rest for now
Pick ONE; practice it deliberately in every response this week
Re-run the same prompts next week; check whether the sting is gone
Then — and only then — move to the second one

Want a trained grader to do the diagnosis pass for you? Get a detailed evaluation with band-level feedback on all four criteria → celpipuni.com/evaluation.

The nuance nobody mentions

Here's my honest position: the "12 mistakes" framing itself is a trap if you read it once and try to avoid everything. Self-monitoring has a bandwidth cost — every sentence you screen against twelve rules is a sentence not spent choosing the better word. The writers I've seen move from 7 to 9 fixed one habit at a time until it disappeared from their writing, then moved on. Automaticity first, breadth later. It feels slow. Six weeks of it is faster than a third retake.

Tools and resources

The official CELPIP site — the current writing tasks, recommended word counts (150–200), and rubric language. Reading the rubric once, slowly, beats most practice tests. (Format gets revised — confirm there.)
Free mock test — instant section scores; take one before and one the week after your fix, and measure: celpipuni.com/free-mock-test
Human evaluation — a trained grader's diagnosis pass is faster than self-diagnosis and catches what your eyes forgive: celpipuni.com/evaluation

Frequently asked questions

Why am I stuck at 7 in CELPIP Writing?

Usually 2–3 compounding habits, not weak English: a buried request, undeveloped reasons, connector-stuffing, or no paragraphing. Each costs a little on a different criterion. Fix them one at a time — a 7 means most of your writing already works.

What's the most common Task 2 mistake?

Sitting on the fence — "both options have merit." The survey asks you to pick one and justify it, so an unmade choice caps task fulfillment regardless of your English.

What's the most common Task 1 mistake?

Burying the request. Three sentences of context before the reader learns what you want. The fix is structural: the request appears by sentence 3, every time.

Are grammar mistakes on this list?

Grammar errors cost points, but at band 7 your grammar is usually clean enough that it's not the capper. The 12 here are strategic and stylistic — they cap accurate writers. For grammar, see CELPIP grammar basics.

Should I fix all 12 at once?

No. Fix one per week. Attempting all twelve overloads your self-monitoring and your fluency drops — the opposite of what you need. One habit, deliberately, until it's gone.

How do I know which mistakes I make?

Write both tasks under timed, typed conditions, then do four reads against this list — one per criterion group. Better: get a human evaluation, because your own errors read as correct to you. That's what makes them errors.

What to do next

Today: write both tasks, timed, typed. Don't edit extra — capture your habits.
Tonight: four passes against this list. Mark your 2–3 stings.
Tomorrow: pick one. Rewrite yesterday's responses fixing only that.
This week: get one evaluation to confirm your diagnosis.
Next 3 weeks: one habit per week. Retest with a mock at the end.

The bottom line

A 7 in CELPIP Writing isn't a verdict on your English — it's a receipt for two or three specific habits. The rubric is public, the mistakes are knowable, and the fix is boring in the best way: find yours, fix one, repeat. Most people don't need a new writing style. They need to stop doing two things.

Find out which mistakes are holding your band down — get a detailed evaluation with band-level feedback → Get your writing diagnosed.