TOEIC Link Writing Task Types and Scoring Criteria: What ETS Actually Counts and What It Quietly Ignores
TOEIC Link Writing is the smallest section by question count and the highest variance section by score per minute. A test taker can move from the middle band to the upper band by changing two habits at the rubric level, and a test taker can lose three score-band steps by misreading the response-length floor on a single task type. The rubric is the lever, not the topic.
This guide breaks down every TOEIC Link Writing task type, the four scoring axes ETS uses to grade each response, and the five rubric details that quietly determine whether the response scores in the upper band or stalls in the middle.
The five Writing task types on TOEIC Link
TOEIC Link Writing rotates five task types across the section. Not every task type appears on every release — typically three of the five appear, with the rotation weighted toward whichever types ETS is currently calibrating. Knowing all five lets you handle the rotation without surprise.
Type 1 — sentence completion from a picture (≈8 minutes total)
The prompt gives you a workplace photograph and a partial sentence with a blank. You complete the sentence to describe what is happening in the photograph. The task tests targeted vocabulary, prepositional accuracy, and tense agreement under a short time floor (typically 60 seconds per item).
The trap: candidates write a full new sentence instead of completing the partial sentence. ETS scores this as a non-response on the relevant rubric axis even if the new sentence is grammatically perfect. The rubric requires you to complete the stem provided, not replace it.
Type 2 — respond to a written request (≈10 minutes total)
The prompt gives you a short workplace email or message (typically 80-120 words) and asks you to write a reply. The reply has a stated purpose and one or two stated constraints (for example: "decline the meeting and propose an alternative time").
The trap: candidates write a reply that addresses the email's topic but skips one of the stated constraints. The rubric treats each constraint as an independent scoring item. Missing one constraint is more costly than weak vocabulary throughout the response.
Type 3 — describe a picture (≈8 minutes total)
The prompt gives you a workplace photograph and asks you to describe it in three to four sentences. The response is graded on vocabulary breadth, sentence variety, and observational completeness — what you noticed in the photograph, not just what was obvious.
The trap: candidates describe only the foreground subject. The upper-band rubric rewards observational layering — foreground subject, background context, inferred relationship between the two. A response that mentions only the foreground will not score above the middle band regardless of grammar quality. For more on how observational layering shows up in Listening Part 1 photo descriptions too, see our TOEIC Link listening strategies by question type guide.
Type 4 — write a short opinion (≈15 minutes total)
The prompt gives you a workplace-relevant question (typically about work practices, communication preferences, or office policy) and asks for an opinion response of 150-180 words. The response is graded on argument structure, supporting examples, and cohesion across paragraphs.
The trap: candidates structure the response as one block of running prose. The upper-band rubric expects three visible structural moves: a position statement, at least two distinct supporting examples or reasons, and a conclusion. A response without these three moves caps at middle-band scoring even with strong vocabulary.
Type 5 — write a structured workplace document (≈15 minutes total)
The prompt asks you to write a workplace document of a specific type — a meeting summary, a project status update, a customer-facing apology, or a process change announcement. The response is graded on document-shape fidelity, tone appropriateness, and content completeness.
The trap: candidates write the document in a tone that does not match the document type. A customer-facing apology written in an internal-memo tone will not score in the upper band regardless of grammatical quality. The rubric checks whether the tone matches the audience the document is addressed to.
The four scoring axes ETS uses on every Writing task
Every Writing task is scored on the same four axes, regardless of task type. Knowing the axes lets you self-audit your practice responses against the actual rubric instead of guessing.
Axis 1 — task fulfillment
Did the response address every stated constraint of the prompt? This is the highest-weight axis on tasks 2, 4, and 5. A response that handles 90 percent of the constraints excellently but skips one constraint entirely will score lower than a response that handles every constraint at competent quality.
Self-audit method: after writing a response, list every constraint stated in the prompt and check each one against the response. If a constraint is not visibly addressed, the response will not score in the upper band.
Axis 2 — language quality
Vocabulary range, grammar accuracy, and sentence variety. This is the most familiar axis but, contrary to common belief, it is not the highest-weighted one on most tasks. ETS weights task fulfillment higher because it is a sharper signal of the test taker's ability to follow workplace instructions.
Self-audit method: count the number of distinct sentence structures in the response. Three or fewer distinct structures cap the language-quality score at the middle band even if every sentence is grammatically perfect.
Axis 3 — organization and cohesion
For tasks 4 and 5 only, the response is graded on whether the structural moves are visible and the transitions between them are smooth. Cohesion vocabulary — however, in addition, consequently, on the other hand — is the dense scoring signal on this axis.
Self-audit method: highlight every cohesion marker in the response. Tasks 4 and 5 responses with fewer than three distinct cohesion markers will not score in the upper band.
Axis 4 — register and tone match
The tone of the response should match the workplace context implied by the prompt. A casual reply to a formal request is penalized; a stiff reply to a casual request is penalized less but still penalized. This axis carries the highest weight on tasks 2 and 5.
Self-audit method: identify the register of the prompt by counting modal verbs (could, would, might) and tentative language (perhaps, it might be worth considering). Match the response's modal density to the prompt's. A response with modal density below the prompt's signals over-directness and will lose on the register axis.
The five rubric details that quietly decide upper-band vs middle-band
These details are not stated explicitly in the public rubric but appear consistently across released sample responses with scoring annotations. They are the bridge between competent middle-band responses and upper-band responses.
Detail 1 — sentence completion vs full sentence
On task type 1, the rubric distinguishes between completing the stem and replacing it. A response that completes the stem with a five-word phrase scores higher than a response that replaces the stem with a perfect ten-word sentence. The rubric reads stem replacement as failure to follow the task instruction.
Operational rule: copy the stem into your response before writing the completion, then verify the completion makes the full sentence grammatical.
Detail 2 — response length floors
Each task type has a response length floor below which the response cannot score in the upper band regardless of quality. The floors are not published explicitly but are observable from sample responses.
- Task 1 — at least 8 words including the stem
- Task 2 — at least 60 words for the reply body
- Task 3 — at least 35 words across the description
- Task 4 — at least 130 words for the opinion response
- Task 5 — at least 110 words for the document body
Under-floor responses are capped at the middle band even if every other axis is at upper-band quality. Over-floor responses are graded on quality, not length.
Detail 3 — off-topic penalty
A response that addresses a related but not-prompted topic is penalized more heavily than a response that addresses the prompted topic incompletely. ETS treats off-topic as a failure of comprehension, which is a separate axis from quality. The penalty applies even if the off-topic response is grammatically excellent.
Operational rule: in the first 30 seconds, write down the exact topic and constraint list from the prompt at the top of your response area, then refer back to it before writing each sentence.
Detail 4 — template signature detection
ETS uses automated scoring assistance on TOEIC Link Writing, and the assistance flags template-signature responses — responses that reuse memorized opening phrases verbatim across multiple tasks. A response that opens with the same six-word memorized phrase as 30 percent of other responses in the same release will trigger a template-signature flag and lose on the language-quality axis.
Operational rule: do not memorize opening phrases. Memorize structural moves and cohesion vocabulary, then assemble the opening sentence from the prompt's specific content.
Detail 5 — CEFR mapping at the boundary
The Writing score is internally mapped to a CEFR band (A2, B1, B2, C1) before being converted to the scaled score. The CEFR boundary at B1-to-B2 is the most consequential — most test takers stall there. The B2 boundary requires three observable features in the response that B1 responses lack: a subordinate clause that depends on a main clause introduced earlier, a hedged claim (using might, could, seems likely), and at least one example or reason that is not the most obvious one a beginner would choose.
For deeper context on how CEFR mapping works across all four TOEIC Link sections, see our TOEIC Link CEFR conversion guide.
A 30-minute practice routine that targets the rubric
The mistake in TOEIC Link Writing prep is practicing in response-time isolation — writing one response, getting a score, and moving on. The rubric rewards specific moves, and the practice routine has to target those moves.
The 30-minute routine that produces band-step gains in two weeks:
- Minutes 0-5 — pick one task type and read three sample upper-band responses for that type, marking the structural moves and cohesion vocabulary
- Minutes 5-15 — write your own response to a new prompt of that task type, with the four scoring axes visible on a sheet of paper
- Minutes 15-22 — self-audit your response against each axis using the methods listed above (constraint check, structure check, cohesion marker count, modal density match)
- Minutes 22-30 — rewrite the weakest one or two sentences in the response, targeting the specific rubric detail that flagged
The point of the rewrite is not to produce a polished final draft; it is to train the recognition pattern that lets you write to the rubric in real time on test day.
What this means for your study time
If you have 20 hours to spend on TOEIC Link Writing, do not spread them evenly across the five task types. Spend 8 hours on task types 4 and 5 (the highest-weight tasks by score impact), 6 hours on task type 2 (the constraint-heavy task most candidates lose points on), and 6 hours across types 1 and 3 combined. This distribution matches the score-weight distribution ETS uses internally, and it concentrates practice on the rubric details that move the band.
The 30-day study plan that operationalizes this distribution is laid out in our TOEIC Link 30-day study plan — the Writing-specific weeks (weeks 3 and 4 of the plan) follow the 8-6-6 distribution above.