Most IELTS Speaking preparation is spent on Part 2. The cue card is the part people fear, the part with the stopwatch, the part every coaching centre drills.
Part 3 is the part that moves your band.
It is the longest section, it has no preparation time, and it is the only part of the test where the examiner can push. If you are sitting between 6 and 7 in Speaking — which is where most candidates in Bangladesh are sitting — the difference is almost always visible in the four or five minutes after the cue card.
What Part 3 actually is
The examiner takes the topic you just spoke about and moves it away from you.
Part 2 was a shop you like going to. Part 3 is: why are small shops closing in your city? Is online shopping good for a community? What do you think high streets will look like in twenty years?
The connection is the topic. The demand is different in three ways:
- No preparation. No minute, no notes, no pen.
- Not about you. The questions are about society, trends, causes, comparisons, the future. Your own life is now only an example you may reach for.
- It develops. The examiner follows your answer. A thin answer invites a harder question, not an easier one.
That last point is the one candidates do not expect. Part 3 is a conversation with a direction, and the direction is decided by what you give them.
The four things being scored, and where Part 3 tests them
The public band descriptors for Speaking have four criteria, each worth a quarter of your Speaking band:
| Criterion | What Part 3 does to it |
|---|---|
| Fluency and coherence | No preparation, so hesitation has nowhere to hide. Discourse markers matter here more than anywhere else in the test. |
| Lexical resource | Abstract topics force abstract vocabulary. You cannot describe "the economy" with the words you used for "my favourite shop". |
| Grammatical range and accuracy | Speculation and comparison pull in conditionals, passives and modals — structures Parts 1 and 2 let you avoid. |
| Pronunciation | Unchanged, but longer stretches of speech give the examiner far more to judge. |
Notice what is not in the table: ideas, opinions, knowledge, being right. Nothing in the descriptors scores the content of your answer. That single fact should change how you prepare, and it almost never does.
Why short answers cap you
Here is a real exchange, lightly tidied, from a practice session:
Examiner: Why do you think people are shopping online more? Candidate: Because it is cheaper and more convenient.
That answer is correct. It is grammatical. It is a 5.5.
Not because it is wrong — because there is nothing in it to score. Nine words give the examiner no evidence of range, no complex structure, no discourse marker, no development. They cannot award what they have not heard.
Now the same idea, developed:
Candidate: Mainly cost and convenience, I think. Prices tend to be lower online because the seller is not paying rent on a shop in a busy area, and that saving usually reaches the customer. The convenience part matters more in a city like Dhaka, though — if getting to a market and back takes two hours in traffic, people will accept a slightly higher price to avoid it. So I would say convenience is actually the stronger reason here, even though cost is the one people mention first.
Same opinion. Same person. The second version contains a causal clause, a comparison, a concession ("though", "even though"), a hedge ("I would say"), and a local example. That is a 7, and nothing in it required a better idea.
The three moves
If you remember one thing, make it this. Every Part 3 answer has three moves:
1. Answer the question. One sentence. Directly. Do not open with "That is a very interesting question" — it buys two seconds and signals a memorised opener.
2. Say why. The reason behind your answer. This is where causal language lives: because, which means, the result is, that is partly why.
3. Add one thing. An example, an exception, a contrast, or a limit on your own claim. Although, that said, in Bangladesh particularly, it depends on.
Three moves, roughly three to five sentences, 30 to 45 seconds. Then stop. Stopping is part of the skill — a candidate who runs on for ninety seconds is not being rewarded for stamina, and the examiner has a fixed amount of time to get through.
Most candidates do move one and stop, because move one answers the question. But the question is only a prompt. What is being measured is the language you build on top of it.
The question types, and what each one wants
Part 3 questions come in recognisable shapes. Knowing the shape tells you which grammar to reach for.
| The question | What it is asking | Language it pulls |
|---|---|---|
| Why do you think…? | Cause | because, due to, which leads to |
| Do you think X will…? | Prediction | will probably, is likely to, I would expect |
| How has X changed? | Comparison over time | present perfect, used to, compared with |
| Is it better to…? | Evaluation | comparatives, whereas, on balance |
| What could be done about…? | Solution | could, should, one option would be |
| Do you agree that…? | Position | concession, while it is true that… |
Two of those are worth special attention. Prediction questions are the single most reliable place in the whole test to produce a conditional, and conditionals are what the descriptors mean by a complex structure. Agreement questions invite a concession — while it is true that X, I would still say Y — which is one sentence that demonstrates range, accuracy and coherence at once.
When you have nothing to say
You will be asked something you have never thought about. Everyone is. It is not a knowledge test, so there is no fact you are missing — but silence scores nothing, and so does "I don't know."
Say the true thing, then reason:
I have not really thought about that before. My first instinct is that it is mostly about cost, because that is usually what decides this kind of thing — although I am not sure that would hold in a smaller town, where people might have fewer options to begin with.
That answer contains an admission, a hedge, a reason, and a qualification. It is better language than a fluent, confident, rehearsed paragraph about something adjacent — and examiners hear the rehearsed paragraph constantly.
What to stop doing
Memorised phrases. It is a double-edged sword. Every coin has two sides. In the modern era of globalisation. Examiners hear these hundreds of times. They do not demonstrate range; they demonstrate preparation, and the descriptors say nothing about preparation.
Opening every answer the same way. "Well, that's a good question" three times running is a tic the examiner will notice.
Trying to be interesting. You are not being scored on insight. A dull, clearly expressed, well-developed answer beats a clever one you cannot build a sentence around.
Padding. Repeating the question back, or restating your point in different words to fill time, is visible and it costs you on coherence. Three moves, then stop.
Answering a Part 1 question. The commonest failure after the cue card is staying personal — "I go shopping with my mother on Fridays" — when the question asked about shopping habits in general. Listen for whether the question says you or people.
How to practise it without a partner
Part 3 is hard to drill alone, but not impossible.
Take any Part 2 cue card. Write three Part 3 questions off it yourself — one cause, one comparison, one prediction. That takes two minutes and is itself useful, because it teaches you the shapes.
Then record yourself answering, with no preparation and no second attempt. Play it back with one question in mind: did I do all three moves? Not "was it good" — did I answer, say why, and add one thing. Most recordings in the first week stop after move one.
Keep going until the three moves are automatic. That is the whole skill. It takes most candidates two or three weeks of ten minutes a day, and it is worth more than another month of cue cards.
The honest summary
Part 3 rewards one thing above all: the ability to keep talking sensibly about something you have not prepared, in sentences that are longer than they strictly need to be.
That is a skill, it is trainable, and it is not the same skill as having good English. Plenty of candidates with strong vocabulary sit at 6.5 because they answer the question and stop. Answer, say why, add one thing — and the band moves without your English changing at all.
