Build the essay. Adapt every paragraph to the question.

There is no skeleton key

In heist films the thief slides one glowing key into the vault and it clicks open. Real locksmiths carry a toolkit: a tension wrench, and the patience to feel out the binding order of the pins in that particular lock.

Most Task 2 preparation is a search for the skeleton key — a memorised template, a list of rare words, a set of transition phrases to glue it together. It does not work, because the examiner is not grading decoration. They are grading disciplined thinking under time pressure, aimed at one exact prompt. You have to read the lock in front of you.

The rules of the game

The Academic Writing test gives you 60 minutes for two tasks. Task 2 presents a point of view, an argument or a problem, and asks for at least 250 words. The usual working split is about 40 minutes on Task 2.

It must be continuous prose — no bullet points, no shorthand — in an academic or neutral register. And it carries twice the weight of Task 1 in your Writing band, which makes it the anchor of the result. A strong graph description will not rescue a weak essay.

The four criteria, and the one people misread

Task Response asks whether you answered the exact prompt and developed ideas that support your position. Coherence and Cohesion asks whether the essay is logically organised and whether the ideas connect. Lexical Resource covers vocabulary. Grammatical Range and Accuracy covers sentence variety and control.

Lexical Resource is the one that gets misread. The common advice is to memorise rare words and deploy them everywhere, but precision beats rarity every time. Rare vocabulary bolted onto a weak or off-topic argument is Ferrari tyres on a bicycle: expensive, conspicuous, and no faster. Relevance and idea development are the foundation; vocabulary decorates a structure that already stands.

Three buckets, so you can tell what is broken

Writing whole essays over and over and staying at Band 5.5 tells you nothing, because you cannot see which part failed. Are you missing the prompt, or are your sentences ungrammatical? Those need completely different exercises.

WeLearn separates preparation into three trainable buckets: the architecture (how you build the response, in order), the question families (how you adapt that architecture to the specific instruction), and the transferable skills (the mechanics operating inside every paragraph). Diagnosis gets much easier once the three are apart.

The architecture, and the four-paragraph caveat

A dependable working sequence: analyse the prompt, write the introduction, write body one, write body two, write the conclusion, then review.

People argue online that IELTS requires four paragraphs. It does not — there is no official paragraph count. Four is a training guardrail that keeps you focused, nothing more.

You can write a third body paragraph, but only when you have a genuinely distinct third idea and the time to develop it properly. Squeezing one in with five minutes left produces a shallow list of leftovers, and two deeply developed paragraphs will always outscore three rushed ones. The test rewards depth of reasoning, not the number of points you can brainstorm.

The instruction, not the topic, sets each paragraph’s job

For planning purposes, prompts are often grouped into five teaching families: opinion, discussion, problem–solution, advantages–disadvantages, and direct questions. Those are teaching categories, not official test terminology.

The rule underneath them is what matters. Take international tourism as a topic. Asked to discuss the advantages and the disadvantages, you are describing a landscape. Asked whether the advantages outweigh the disadvantages, everything changes: your thesis now needs a definitive verdict, you have to take a side, and the body paragraphs have to weigh the two against each other rather than list them.

Write the first essay in answer to the second prompt and you can produce something grammatically clean that still damages your Task Response score — because you brought the wrong tool for that lock.

Paraphrasing is five mechanical moves

"Say it in your own words" is the most common advice about paraphrasing and the least useful. There are five concrete moves: swap synonyms, change word order, change word form (a noun into an adjective), change the voice, or restructure the sentence.

Relying only on synonym swapping is where it goes wrong. Turn "the government spends money on space exploration" into "regimes squander currency on cosmic voyages" and you have changed the meaning: "squander" asserts waste, which the prompt never claimed, and "regimes" carries a political charge that "government" does not.

A paraphrase that reads beautifully but shifts the logic of the prompt costs you marks. Changing the grammatical structure — "spending money" becoming "the allocation of funds" — is far safer than hunting for impressive synonyms.

A thesis has to be arguable

If no reasonable person could disagree with your position, you have not stated a position — you have stated a fact. The position then needs to be visible in the introduction, defended through the body, and restated in the conclusion.

Avoid previewing your exact reasons in the thesis ("I agree for two reasons, cost and time"). It locks your body paragraphs into a promise made before you had time to think, and leaves nothing for them to develop.

Build a chain, not a list

This is the single most common reason candidates stall at Band 6. Asked about public transport, the instinct is to show breadth: it is cheap, it reduces pollution, it eases congestion, it is safer than driving. Four true claims — and zero reasoning. A pile of stones is not a bridge.

Instead, make one claim and keep asking what follows from it. "Public transport offers a significantly cheaper alternative to private vehicle ownership." So what? "Because the cost per journey is lower, it puts daily mobility within financial reach of lower-income households." So what? "That mobility grants access to employment and education that would otherwise be out of reach."

Three linked sentences demonstrate the ability to construct an argument. Five isolated claims demonstrate that you can remember facts.

Do not invent evidence

Candidates write things like "a 2023 Oxford study found that 83% of people prefer buses". The examiner knows there is no academic database in the exam room, and the invented precision helps nothing.

Ground an abstract claim by explaining the mechanism instead: in densely populated cities, dedicated bus lanes let commuters bypass rush-hour traffic, which is what makes the option efficient. Illustrate how it works rather than fabricating how many.

Connectors are claims about logic

Choose linking words by their function — addition, contrast, cause and effect — not by which ones you have not used yet. A connector asserts a relationship between two ideas, so "furthermore" belongs only where you are genuinely adding a parallel point.

The most valuable function is concession: acknowledging a real point from the other side before showing your position still holds. "Despite the high initial cost of building a subway system, the long-term economic benefits far outweigh the initial debt." That shows you examined the whole topic rather than only your half.

One grammar trap lives here. "Although the scheme was expensive, but it went ahead" announces the contrast twice. "Although" has already done the work; the "but" is redundant. Write "Although the scheme was expensive, it went ahead."

Sentence variety with a purpose, and a real review

Mix simple, compound and complex sentences for a reason rather than for the mixture. A short sentence lands a concluding thought; a complex one carries cause and effect.

The final review is your last line of defence: check that you answered the prompt, that the logic chains hold, that your position has not drifted, and that basic language errors are caught.

The 5/30/5 clock

A practical way to spend the 40 minutes: about five minutes decoding the prompt and assigning each paragraph its job, about 30 minutes writing, and the final five on the review. This is a WeLearn practice target, not an official IELTS allocation.

The scenario worth rehearsing: you reach the review at 230 words with three minutes left. Do not add filler — empty summarising sentences damage Coherence and Cohesion immediately. Go back to the least developed point in your argument and add one more link to its chain. Ask what follows from it one more time and write the outcome. You clear the word count and improve Task Response in the same move.

A weekly loop that is not just writing more essays

A timed 40-minute essay every day is exhausting and inefficient. This cycle is more surgical.

• Write one timed baseline essay to see where you stand.
• Diagnose it four times, once through each official criterion. Read it asking only "did I answer the prompt?", then only for how sentences connect, then only for vocabulary, then only for grammar.
• Pick one weak sub-skill from that diagnosis — say, clumsy concession sentences.
• Drill that skill alone. Pull five prompts from different families and write five concession sentences. You do not fix a free throw by playing another match.
• Rewrite the section of your baseline essay that contained the error, using the sharpened skill.
• Transfer it into a new full timed essay.
• Keep an error log, so next week you know exactly what to drill.

The takeaway

Task 2 is not a memory test. Nobody is looking for a candidate who swallowed a dictionary or memorised a one-size-fits-all template.

What it measures is the ability to read the exact instruction in front of you, choose a defensible position, give every paragraph a specific job, and develop it in clear chains of reasoning. You do not need a skeleton key when you can read the lock.