How Part 3 works, and how this page differs from our Part 3 guide
IELTS Speaking Part 3 is a four to five minute discussion in which the examiner asks you about issues "related to the topic in Part 2 in a more general and abstract way and, where appropriate, in greater depth", per the ielts.org Speaking format page. The way to answer any Part 3 question is the same four moves: a direct position in the first sentence, the strongest reason with one concrete example, a limit or a second view, and a one-sentence resolution. Four to six sentences, roughly 30 to 60 seconds, then stop and let the examiner push back. That is the whole method; the rest of this page shows how the four moves bend for each of the six question types the examiner uses, and demonstrates them on ten questions with model answers written by TalkDrill to the Band 8 descriptor lines.
We already have a general guide, IELTS Speaking Part 3 questions and answers, which covers common topics with sample answers, and the sub-hub article Part 3 discussion topics, which lists topics by theme. Start there if Part 3 is new. This page is the structural version: it sorts every Part 3 question into a type, gives you a frame per type that works on any topic, pairs each of the ten bank questions with the pushback the examiner is likely to add, and explains how the examiner scores what you say. The ten questions are the same ten in our practice bank, so you can read them here and then argue them with the AI examiner, who will push back the same way.
- Four to five minutes, the last part of a test of 11 to 14 minutes. In our experience, five to eight questions; IELTS does not publish a count.
- Topic set by your Part 2 card, moved from you to people in general. "Describe a quiet place" becomes "Is it getting harder to find quiet places in cities?"
- A conversation, not a monologue: the examiner follows up, challenges and redirects. That is designed in, not a sign you answered badly.
- No separate score. The four criteria are applied to "average performance across all parts of the test" (note ii, descriptors PDF).
- You can ask for a question to be repeated or put another way; do not expect the examiner to give their own opinion or to tell you whether yours is right.
The six question types (and two variants)
Every Part 3 question we have collected in the current format is one of six types, or a variant of one. Recognising the type in the first second tells you which frame to use, which tense to lead with and what the pushback will be, before you have thought about the content. The table is ours; IELTS does not publish a typology, but the wording of the questions is consistent enough that the sort is reliable.
| Type | How the question is worded | Tense and language it wants | The pushback that usually follows | In this set |
|---|---|---|---|---|
| Opinion | "Should...?", "Do you think...?", "Is it important that...?" | Present simple; position language ("I think", "on the whole", "within limits") | The opposite view, or "is that not someone else's job?" | Questions 1, 7 |
| Compare | "How is X different from Y?", "Which is better...?", "Do men and women...?" | Comparatives, "whereas", "on the other hand", criteria named | "So is Y worse?" (forces you to rank) | Question 2 |
| Predict | "How will... change in the future?", "Do you think people will still...?" | "Will", "is likely to", "I would expect", hedged certainty | "Is that positive or negative?" | Question 3 |
| Cause | "Why do people...?", "What makes some people...?" | Causal language ranked: "the main reason", "a second factor", "partly because" | A simpler cause offered for you to accept or reject | Questions 4, 9 |
| Solution | "What could governments / schools / parents do about...?" | Modals ("could", "should", "would need to"), policy nouns, costs admitted | "Would people accept that?" | Question 5 |
| Evaluate | "Is X a good thing?", "Is X harmful?", "What are the advantages and disadvantages of...?" | Weighing language: "on balance", "the clearest benefit", "the cost is" | "Can it be controlled?", "Does that apply everywhere?" | Question 6 |
| Variant: past versus now | "How has... changed compared with your parents' generation?" | Present perfect for the change, past simple for then, present for now | "Has something been lost?" | Question 8 |
| Variant: hypothetical | "If everyone..., what would happen?", "What would life be like without...?" | Second conditional sustained; "would" throughout | "Would you want that?" | Question 10 |
The two variants are worth separating because they are grammar traps rather than content traps. A past-versus-now question is answered in three tenses, and Hindi, Punjabi and Vietnamese speakers who narrate everything in the present lose the Band 7 line "error-free sentences are frequent" in the first twenty seconds. A hypothetical question is answered in the second conditional from start to finish; most candidates open correctly with "would" and drift into "will" by the second sentence. Both patterns are shown in questions 8 and 10 below.
A framework for each type
One frame underlies all six: position, reason with example, limit, resolution. What changes by type is what each move contains and which words carry it. The examples under each frame are written by TalkDrill as single moves, not full answers; the full answers are in the next section.
Opinion: position, strongest reason, concession, resolution
Say where you stand in the first sentence, with a qualifier that is true ("within limits", "on the whole", "in most cases"). Give the strongest reason, not the first reason, and attach one concrete example to it. Concede the strongest point against you in one sentence. Resolve: say how both can be true. Example moves: "I think they should, within limits." / "The strongest argument is that these are skills every adult needs, whereas a lot of what we memorise at school is forgotten within a year." / "The limit is that timetables are already crowded." / "So I would teach them through existing subjects rather than as another exam." Under pushback, keep the position and accept part of the objection: "You are right that families do most of it, but not every family can."
Compare: name the criteria, compare on each, then match to situations
Do not compare in general; compare on two or three named criteria (cost, feedback, access, speed). State which side wins each, then close by saying which option suits which situation, because "it depends" is only a Band 8 answer when you say what it depends on. Example moves: "The main difference is feedback." / "A person corrects you on the spot, whereas a video leaves you to find your own mistakes." / "Online wins on access and pace." / "So they suit different stages: online to get started, a person to get good."
Predict: a visible cause, the change it will produce, the certainty you really have
A prediction is credible when it starts from something already happening. Name the cause you can see today, say what it will produce, and vary your certainty honestly: "will" for what is already in motion, "is likely to" for the probable, "I would expect" for your own judgement. End with the open question. Example moves: "Remote work has already shown people that a job in Bengaluru does not require living in Bengaluru." / "So I think second-tier cities will grow much faster." / "Whether housing keeps up is the real question."
Cause: rank the causes, give each a mechanism, name the weakest last
Three causes are enough, and ranking them is what separates a list from an analysis. For each, say how it works in one sentence, not just that it exists. Put the cause the examiner expects you to name (willpower, laziness, money) last, and say why it matters less than people think. Example moves: "There are three causes, and willpower is the least important." / "The first is that the reward is delayed: walk for a month and nothing visible happens." / "The second is environment: there is nowhere safe to walk." / "The third is that people start too big."
Solution: sequence the measures, add one local one, admit the cost
Solutions sound naive when they ignore order and cost. Say what must happen first (usually making the alternative better) and what can follow (making the problem costlier). Add one measure specific to your country, because that shows you are thinking rather than reciting. Then admit what it costs and who pays, which is where "would people accept it?" is already answered. Example moves: "The most effective single step is to make the alternative better before making driving worse." / "Once that exists, a government can charge for driving into the centre." / "In India I would add staggered office hours." / "None of this is cheap, but the cost of the traffic is already being paid, just invisibly."
Evaluate: verdict, the reason it rests on, the clearest case, the exception, the fair position
Evaluation is weighing, so the language is "on balance", "the clearest harm is", "I would not say all", "the fair position is". Give a verdict in the first sentence and say which single reason it rests on. Give the clearest case for it, then the exception that stops it being absolute, then state the position that survives both. Leave the hardest part named but open; a Band 8 answer does not pretend to have solved online advertising. Example moves: "On balance, yes, and for a specific reason: a child cannot yet tell information from persuasion." / "The clearest harm is in food." / "I would not say all advertising to children is harmful; a hand-washing campaign is advertising too." / "So the fair position is that commercial advertising to young children should be restricted."
The 10 questions with Band 8 model answers
Each question below carries its type, the framework the model uses, the examiner's likely pushback and a model answer written by TalkDrill to the Band 8 descriptor lines: "topic development is coherent, appropriate and relevant", "wide resource, readily and flexibly used to discuss all topics and convey precise meaning", "wide range of structures, flexibly used", and hesitation that is "content related". The models are 110 to 130 words, which is about 45 to 60 seconds spoken at a natural pace, and they are deliberately not neutral: each one takes a position and defends it, because that is what the pushback is for. The topics continue the cue-card families on our cue cards page (a person who taught you a skill leads to questions 1 and 2; a place that has changed leads to question 3; a healthy habit leads to question 4; an advertisement leads to question 6), which is how a real Part 3 follows a real Part 2.
1. Opinion: "Should schools teach practical skills such as cooking and managing money?"
Framework used: Position, strongest reason with an example, the limit, the resolution. Examiner's pushback: "But is that not the job of the family rather than the school?"
Band 8 model answer (written by TalkDrill)
I think they should, within limits. The strongest argument is that these are skills every adult needs, whereas a lot of what we memorise at school is forgotten within a year. In India, for example, most people learn about tax, interest and insurance only when they make their first expensive mistake. A single weekly class from the age of fourteen would prevent a lot of that. The limit is that schools already have crowded timetables, so I would not want cooking to replace science or languages. I would rather see practical skills taught through the existing subjects, for instance teaching percentages through a real household budget, than added as another separate exam.
Examiner note. Fluency and coherence. The Band 8 opinion answer states a position, gives the strongest reason with one concrete example, concedes a limit and then resolves it. The pushback tests whether you can hold the position while accepting part of the examiner's point, not whether you change your mind.
What to copy. The pushback ("is that not the family's job?") is answered by accepting part of it and keeping the position.
2. Compare: "How is learning a skill from an older person different from learning it online?"
Framework used: Name the criteria (feedback, judgement, access, pace), compare on each, then say which suits which situation. Examiner's pushback: "So is online learning worse, in your view?"
Band 8 model answer (written by TalkDrill)
The main difference is feedback. An older person watching you ride a bike or cook a dish can see exactly what you are doing wrong and correct it on the spot, whereas a video shows you the right way and leaves you to find your own mistakes. A second difference is that a person passes on judgement as well as technique: my uncle taught me when not to overtake, not just how. Online learning wins on access and pace. You can watch a step ten times at midnight without embarrassing anyone, and you can learn things nobody in your family knows. So I would say they suit different stages: online to get started, a person to get good.
Examiner note. Lexical resource and coherence in comparison. The examiner wants comparative language ("whereas", "wins on", "suit different stages") and clear criteria for the comparison (feedback, judgement, access, pace). Answering only about yourself, "I like learning from my uncle", misses the general question.
What to copy. Comparative language throughout: "whereas", "wins on", "suit different stages".
3. Predict: "How do you think cities in your country will change over the next twenty years?"
Framework used: Two shifts, each with a cause already visible today, then the open question. Examiner's pushback: "Do you see those changes as positive or negative?"
Band 8 model answer (written by TalkDrill)
I expect two big shifts. The first is that the growth will move away from the largest cities. Mumbai and Delhi are close to the limit of what their roads and water supply can take, and remote work has already shown people that a job in Bengaluru does not require living in Bengaluru. So I think the second-tier cities, places like Indore, Kochi or Coimbatore, will grow much faster. The second shift is in transport: metro systems are being built in city after city right now, and in twenty years I would expect most middle-class commuting to be on rail or electric vehicles rather than petrol scooters. Whether housing keeps up is the real question, and history suggests it will not.
Examiner note. Grammatical range and accuracy, in future forms. A Band 8 prediction varies its certainty: "I expect", "I think ... will", "I would expect", "the real question is whether". Saying "will" for everything sounds like a forecast you cannot support; hedging every sentence sounds evasive.
What to copy. Certainty is varied: "I expect", "I think ... will", "I would expect", "the real question is whether".
4. Cause: "Why do many people find it hard to keep up healthy habits?"
Framework used: Three causes ranked, each with a mechanism, the weakest named last. Examiner's pushback: "Is it not simply a lack of willpower?"
Band 8 model answer (written by TalkDrill)
I think there are three causes, and willpower is the least important of them. The first is that the reward is delayed. If you walk every morning for a month, nothing visible happens, whereas skipping it gives you an extra thirty minutes in bed today. Humans are not built to value a benefit that arrives in a year. The second is environment. In most Indian cities there is nowhere safe to walk, and the cheapest food on every corner is fried. The third is that people start too big; they join a gym, buy the clothes, go five times, and quit. The people who succeed usually start with something so small that stopping would feel silly.
Examiner note. Fluency and coherence. Listing causes is easy; ranking them and defending the ranking under pushback is the Band 8 skill. The examiner's "is it not simply willpower" is an invitation to disagree politely with evidence, which is exactly what the model does.
What to copy. The pushback offers a lazy cause (willpower); the model disagrees politely with evidence.
5. Solution: "What could governments do to reduce traffic in large cities?"
Framework used: Order the measures (make the alternative better first, then make the problem costlier), add one local measure, name the cost honestly. Examiner's pushback: "Would people in your country accept measures like that?"
Band 8 model answer (written by TalkDrill)
The measures that work elsewhere are well known; the difficulty is political, not technical. The most effective single step is to make the alternative better before making driving worse: frequent buses with their own lanes, metro lines that reach the suburbs where people actually live, and safe footpaths for the last kilometre. Once that exists, a government can charge for driving into the centre, as London and Singapore do, and raise parking prices so that a car is a choice rather than the default. In India I would add staggered office hours, because half the problem is that every office opens at nine thirty. None of this is cheap, but the cost of the traffic itself, in fuel, time and health, is already being paid, just invisibly.
Examiner note. Lexical resource, specifically the vocabulary of policy: "measures", "staggered office hours", "charge for driving", "the default". The pushback tests register; a Band 8 candidate answers "would people accept it" with a realistic view rather than "yes, if the government explains properly".
What to copy. Policy vocabulary: "measures", "staggered office hours", "the default".
6. Evaluate: "Is advertising that targets children harmful?"
Framework used: Verdict with the reason it rests on, the clearest case, the exception, the fair position, the unresolved part. Examiner's pushback: "Children see advertisements everywhere now. Can it really be controlled?"
Band 8 model answer (written by TalkDrill)
On balance, yes, and more harmful than advertising aimed at adults, for a specific reason: a child cannot yet tell the difference between information and persuasion. An adult watching a car advert knows it is trying to sell something; a six-year-old watching a cartoon character eat a particular cereal does not. The clearest harm is in food, where the products advertised most heavily to children are almost always the ones a doctor would limit. I would not say all advertising to children is harmful; a public campaign about washing hands is advertising too. So the fair position is that commercial advertising to young children should be restricted, as some countries already do on television. Doing the same online is much harder, and that is where the real problem now sits.
Examiner note. Fluency and coherence in evaluation. "On balance", "for a specific reason", "I would not say all", "the fair position is": the examiner hears a candidate weighing, not just asserting. The pushback about online control asks for a concession plus a view, not a solution.
What to copy. "On balance", "I would not say all", "the fair position is": weighing, not asserting.
7. Opinion (with a distinction): "Do people rely too much on technology to solve everyday problems?"
Framework used: Split the question into the part that is true and the part that is exaggerated, then draw the line precisely. Examiner's pushback: "If it works, why is that a problem?"
Band 8 model answer (written by TalkDrill)
In some areas yes, in others I think the worry is exaggerated. Nobody complains that we rely too much on electricity, and in ten years I suspect map apps will feel the same; the skill of reading a paper map is gone and almost nobody misses it. Where I do think the reliance is unhealthy is in thinking and remembering. If every small question is answered by a search before you have tried to work it out, you lose the habit of working things out, and that habit matters when the phone is not there, in an exam, an interview or an emergency. So my answer is that the problem is not dependence on technology for tasks but dependence on it for judgement.
Examiner note. Lexical resource. The Band 8 move is the precise distinction at the end: "dependence for tasks" versus "dependence for judgement". That single contrast shows "precise meaning" in the Band 8 descriptor far better than any idiom would.
What to copy. The distinction "dependence for tasks" versus "dependence for judgement" is Band 8 precision.
8. Compare across time (past versus now): "How has the way people celebrate festivals changed compared with your grandparents' generation?"
Framework used: Three changes in present perfect, the past in past simple, then a loss and a gain. Examiner's pushback: "Do you think something has been lost?"
Band 8 model answer (written by TalkDrill)
Three things have changed, in my view. Festivals have become shorter, because people have fewer days off and families are spread across cities, so a celebration that once took a week is now compressed into a weekend. They have become more commercial; my grandmother made sweets at home for Diwali, and my mother orders them, and the shops now start advertising a month in advance. And they have become more visible, in the sense that a lot of the celebration is now performed for photographs. Something has been lost, yes, mainly the slow preparation that brought the family together before the day itself. But something has been gained too: my cousins abroad join on a video call, which my grandparents could never have imagined.
Examiner note. Grammatical range and accuracy: present perfect for change over time ("have become shorter"), past simple for the grandparents, present simple for now. Hindi and Punjabi speakers often flatten all three into the present; the examiner hears that immediately in a past-versus-now question.
What to copy. Tense discipline: "have become shorter", "my grandmother made", "my mother orders".
9. Cause, then evaluate under pushback: "Why do some people stay in jobs they dislike?"
Framework used: Causes that are rational from inside, then the blunt pushback answered directly: "just leave ignores all three". Examiner's pushback: "Should they just leave?"
Band 8 model answer (written by TalkDrill)
Mostly for reasons that are rational, even if they look like weakness from outside. The obvious one is money: in a country without much of a safety net, a disliked salary beats no salary, and a person with a home loan and parents to support cannot afford a gap. The second is identity; people have told their families and neighbours what they do, and leaving feels like admitting failure. The third is simply not knowing what else they could do, because most of us were never shown the range of jobs that exist. "Just leave" ignores all three. The better advice is to leave slowly: build the next skill, save six months of expenses, and then go.
Examiner note. Fluency and coherence. The pushback is deliberately blunt; the examiner wants to see whether you can disagree with the premise of a question without becoming defensive or vague. "Just leave ignores all three" is a direct answer that still respects the question.
What to copy. Disagreeing with the premise without becoming defensive.
10. Predict (hypothetical): "If most people worked from home, how would society change?"
Framework used: Second conditional sustained: three consequences (geographic, social, fairness) and a balanced close. Examiner's pushback: "Would you want to live in that society?"
Band 8 model answer (written by TalkDrill)
The first change would be geographic. If the office stopped mattering, the reason to pay Mumbai rents would weaken, and I would expect a steady movement to smaller towns, with the consequences that follow: property prices falling in the centre, rising in places that were cheap, and local shops in those towns doing well. The second would be social. Offices are where many adults make their friends and meet their partners; without them, people would need to build that elsewhere, and some would not manage it. The third is fairness. Work from home suits people with a quiet room and a good connection, which in most Indian homes is not the situation. So I think it would make life better for one group and lonelier and more unequal for another.
Examiner note. Grammatical range and accuracy: the second conditional sustained across a long answer ("if the office stopped mattering ... I would expect"). Candidates who open in the conditional and drift into "will" within two sentences lose the "range of structures flexibly used" line.
What to copy. Stays in "would" throughout; drifting into "will" is the common slip.
How Part 3 is scored
There is no Part 3 score. The examiner applies the same four criteria, Fluency and Coherence, Lexical Resource, Grammatical Range and Accuracy, and Pronunciation, to the whole test, and the notes at the foot of the official Speaking Band Descriptors PDF say two things that matter for Part 3 in particular: a candidate "must fully fit the positive features of the descriptor at a particular level", and is "rated on their average performance across all parts of the test". So Part 3 cannot be skipped or coasted, and it cannot be rescued by one brilliant sentence either; what counts is whether the Band 7 or Band 8 lines describe you across the four or five minutes. What Part 3 does, more than the other parts, is give the examiner evidence for the top of each criterion, because abstract questions demand the language the Band 8 lines name.
| Criterion | Band 7 line from the PDF | Band 8 line from the PDF | What it looks like in Part 3 (our reading) |
|---|---|---|---|
| Fluency and Coherence | "Able to keep going and readily produce long turns without noticeable effort." Hesitation "will not affect coherence". | "Fluent with only very occasional repetition or self-correction." Hesitation "may occasionally be used to find words or grammar, but most will be content related." | Pauses fall between moves, not inside sentences. You can say "let me think" before a position, but not "the... the... the" in the middle of the reason. |
| Lexical Resource | "Resource flexibly used to discuss a variety of topics." "Some ability to use less common and idiomatic items and an awareness of style and collocation." | "Wide resource, readily and flexibly used to discuss all topics and convey precise meaning." "Skilful use of less common and idiomatic items despite occasional inaccuracies." | Precise distinctions ("tasks" versus "judgement"), policy and social nouns ("measures", "safety net", "the default"), paraphrase when a word is missing. |
| Grammatical Range and Accuracy | "A range of structures flexibly used. Error-free sentences are frequent." "A few basic errors persist." | "Wide range of structures, flexibly used. The majority of sentences are error free." "A few basic errors may persist." | Conditionals sustained, present perfect for change, passive for policy ("could be restricted"), relative clauses carrying examples. Note that even Band 8 allows a few basic errors. |
| Pronunciation | All the positive features of Band 6 and some of Band 8. | "Can sustain appropriate rhythm. Flexible use of stress and intonation across long utterances." "Can be easily understood throughout." | Stress on the contrast word ("not dependence on technology for tasks but for judgement"), a falling tone to close a position, a rise to concede. |
Two consequences follow. First, a short answer in Part 3 is the most expensive short answer in the test, because the examiner has come to Part 3 looking for the Band 7 and 8 evidence that Part 1 cannot provide; "yes, I think so, because it is important" leaves the top half of every criterion unproven. Second, perfect agreement is a trap. If you agree with every pushback, your answers get shorter and your structures get simpler. The examiner's "is it not simply willpower?" is an invitation to use "I would not put it that way, because", which is a Band 8 structure in a Band 8 register. For the full line-by-line translation of the descriptors, including the same answer rewritten at Bands 5 to 8, read our band descriptors page.
The five errors that cap Part 3 at Band 6
These are the patterns we hear most often in Part 3 recordings from learners who are otherwise at Band 7 in Parts 1 and 2. Each maps to a descriptor line; the mapping is our reading, the lines are official.
| Error | What it sounds like | Descriptor line it triggers | Fix |
|---|---|---|---|
| The one-line answer | "Yes, I think schools should teach it. It is very useful." | No evidence of "long turns" (Band 7 Fluency). The examiner marks what was heard. | Four moves, every time: position, reason with example, limit, resolution. |
| The memorised opener | "That is a very interesting question. Well, there are many points of view on this issue..." | "Apparently memorised utterances" (Band 3 grammar note) and overused markers (Band 5 Fluency). | Start with the position. If you need a second, say "let me think" and then start with the position. |
| The personal answer to a general question | "In my case, I like learning from my uncle." | Topic development not "relevant" (Band 8 Fluency); resource not shown on "all topics" (Band 8 Lexical). | Answer about people in general first; use yourself as the example, introduced with "for instance, in my own family". |
| The tense collapse | "Before, people are celebrating for one week, now they celebrate one day." | Band 6 grammar: "errors frequently occur in complex structures". Past-versus-now questions expose it immediately. | Present perfect for the change, past simple for then, present simple for now. Say the three tenses aloud before you answer. |
| The idiom in the wrong place | "Advertising to children is harmful, it is a double-edged sword, every coin has two sides." | "Inaccuracies in word choice and collocation" (Band 8 Lexical); inappropriacy (Band 7). | Replace the idiom with a precise distinction. "Harmful in food, useful in public health" is worth more than any proverb. |
A sixth habit is specific to Indian English and worth naming: "basically", "actually" and "only" as fillers ("basically, people are only lazy, actually"). The Band 5 Fluency line is "overuse of certain discourse markers"; three of these in one answer is the pattern. Our vocabulary and idioms page has a filler audit and the discourse markers that replace them, sorted by the move they belong to (position, reason, concession, resolution).
A practice route for Part 3
Part 3 improves fastest when you practise the frames on topics you have not seen, because the whole skill is building an answer in the three seconds after an unfamiliar question. Ten days is enough to make the four moves automatic.
- Days 1 to 2: sort before you speak. Read the ten questions above and, before reading any model, say the type aloud and the first sentence you would use. Then read the model and compare your opening move with it. Do not compare vocabulary yet.
- Days 3 to 6: one type a day, two questions each. Opinion (questions 1 and 7), compare and past-versus-now (2 and 8), predict and hypothetical (3 and 10), cause and solution (4, 5 and 9), evaluate (6). Answer cold, recorded, 30 to 60 seconds. Then answer the pushback without repeating anything you said the first time. Listen back for the four moves.
- Day 7: the pushback drill. Have a partner, or the TalkDrill AI examiner, ask only the pushback questions from the set, with no first question. Your answer must open with a position that accepts part of the challenge and keeps the view: "You are right that..., but...". The full IELTS Speaking mock runs Part 3 straight after your cue card, on the same topic, with follow-ups.
- Day 8: new topics. The topics in the test are not these ten. Take any abstract topic generator, including our abstract topic generator, and apply the frames to five topics you have never discussed. If you want a live human to argue with, the group discussion rooms are Part 3 at conversation speed.
- Day 9: grammar pass. Listen to your Day 3 to 6 recordings for the two variant traps: did the past-versus-now answer use three tenses, and did the hypothetical stay in "would". Re-record those two.
- Day 10: a full test. Parts 1, 2 and 3 in sequence, timed, on the mock. Score yourself on the four criteria across the whole test, as the examiner does, using the checklist on the band descriptors page. If you are aiming at 8, the Band 8 guide says what to add after the frames are automatic.
The practical frame around all this: in India the test is booked through IDP and costs INR 19,000 per the IDP IELTS India fee page (read 20 October 2026). If Speaking alone misses your target, a One Skill Retake can be booked for INR 12,650 and must be sat within 60 days of the original computer-delivered test, once per test, per the ielts.org One Skill Retake page; our results and One Skill Retake page covers who accepts it. Part 1 and Part 2 have their own pages, Part 1 questions 2026 and cue cards 2026, and the IELTS Speaking hub holds the rest.