Skimming vs Scanning in IELTS Reading: When Each One Is the Wrong Choice

Almost every IELTS Reading guide says the same two things. Skim the passage for the general idea, then scan for keywords to find each answer. Both techniques are real. Both are taught as if they were always the right move, and that is where the marks go.

Scanning is the fastest way to find the wrong answer to a True, False, Not Given question. Skimming is the fastest way to pick the wrong heading. Neither technique is broken. They are being applied to question types that punish them.

This guide is about the choice, not the mechanics: when each one is wrong, what to use instead, and the timing arithmetic that decides whether you finish. The worked examples all come from our own reading bank, so you can go and try them.

#Three gears, not two

Two techniques is one too few. Reading well under time means moving between three.

• Skimming answers "what is this about and where does each idea live". You read titles, first sentences, and the odd last sentence. Speed matters more than understanding.
• Scanning answers "where is this exact thing". Your eye is hunting a shape on the page: a capital letter, a numeral, a year, an unusual word. You are not reading at all.
• Close reading answers "what does this actually say". Two or three sentences, slowly, including the sentence before and the sentence after. This is the gear that earns the mark.

Most candidates have gears one and two and no gear three, so they locate the right region and then still answer wrongly. A smaller group has only gear three, reads everything carefully, and runs out of time at question 30.

Here is the arithmetic that kills gear three used alone. The three passages in our RT021 Academic mock run to 2,514 words. At 200 words a minute, reading them once end to end takes about 13 minutes. At 150 words a minute, which is a realistic pace on unfamiliar academic prose in a second language, it takes about 17 minutes. That is a quarter of the test spent before answering a single question, and you will still have to go back into the text for nearly every one.

#When scanning is the wrong choice

Scanning works when the question contains an anchor: something that appears in the passage in a form your eye can catch and that appears only once or twice. A year. A number. A proper noun. A technical term.

Scanning fails, badly and confidently, when the question's real content is a relationship, an attitude, a quantity word or an absence. In those cases the keyword you scan for is present in the passage and the answer is not.

#The trap in one question

From Passage 1 of RT021, on the marine chronometer. The statement is:

"Larcum Kendall produced his copies of H4 with Harrison's personal assistance."

Scan for "Kendall" and you find him in about two seconds, in a sentence saying he constructed three copies of H4 between 1769 and 1774. Everything in the statement seems to be there. Kendall, copies, H4. The answer is NOT GIVEN, because the passage never says whether Harrison helped.

Scanning did its job perfectly and handed you a wrong answer, because the load-bearing part of the statement was "with Harrison's personal assistance", and that phrase has no anchor to scan for. Absence cannot be found by looking.

Two more from the same test, both NOT GIVEN:

• "John Harrison was originally trained as a clockmaker by his father." The passage calls him self-taught. It says nothing about a father, so this is not FALSE, it is NOT GIVEN. Candidates who scan "Harrison" get forty hits and choose based on the nearest plausible sentence.
• "Behavioural economists are funded primarily by central banks", from the Passage 3 Yes, No, Not Given set. There is no anchor at all. Nothing in that statement appears in the passage. Candidates spend ninety seconds scanning for a word that is not there, and that is the real cost.

The rule: before you scan, ask what you are scanning for. If you cannot name a single word that must be printed in the passage, stop. You are not looking for information, you are looking for its absence, and thirty seconds of failed searching is your evidence for NOT GIVEN, not a reason to keep hunting.

#Quantity words and absolutes

Another RT021 statement, this one FALSE:

"Nudges alone can fully solve the problem of low household saving."

The passage says supporters of the idea themselves treat nudges as "a relatively inexpensive complement to, not a replacement for, broader policy". Scanning "nudges" lands you in the right paragraph immediately. The answer then turns entirely on two small words in the question, alone and fully, which have no counterpart to scan for.

Our trap-rich reading set was built around exactly this. It holds 16 passages and 208 questions, and every explanation is written in the form WHY, then TRAP, then TECHNIQUE, with the trap named. The named types are Paraphrase, Quantifier Misdirection, Time or Detail Flip, MCQ Partial-Truth Distractor, Hedged versus Definitive, Distractor Named Entity, Opinion Attribution, and Negation or Contrast. Look at that list and notice how many of them are invisible to scanning. Six of the eight turn on a word that carries no anchor: a hedge, a quantifier, a negative, or an attribution.

#When scanning is exactly right

The opposite case deserves as much attention, because candidates who get burned by NOT GIVEN start close reading everything and then run out of time.

Scan without hesitation when the question hands you a unique printed thing:

• Sentence completion. "The British Parliament passed the Longitude Act in the year _______." Scan for "Longitude Act". One hit. Answer in fifteen seconds.
• Numbers. "On the Jamaica voyage, H4 lost only _______ seconds over eighty-one days." Scan for the numeral, not the words.
• Matching features with named people. RT021 asks you to match Fred Urquhart, Steven Reppert and Worcester Polytechnic Institute to descriptions. Three proper nouns, each printed once. This is the cheapest group of marks in the whole test and many candidates leave it until they have no time left.

One warning on names. The reading bank tags a whole trap category Distractor Named Entity, because a passage that discusses four researchers will attach the wrong view to the wrong name if you scan for the name and read only the sentence around it. Scan to locate the name, then close read to check who is claiming what.

#When skimming is the wrong choice

Skimming means reading first sentences. Matching headings is the question type where that advice is repeated most often, and it is also where it fails most often, because topic sentences are not reliably first.

Passage 2 of RT021 is about monarch butterfly migration, seven paragraphs, ten headings. Take paragraph C. Its first sentence is:

"Subsequent research has largely revealed that monarchs employ a sophisticated suite of navigational mechanisms."

Skim that and you have "navigation". Now look at the heading list. Two headings are about navigation: "The clock that lies in the antennae" and "A backup system based on magnetism". The first sentence cannot separate them. It is a signpost sentence, not a topic sentence.

The paragraph's actual subject arrives halfway down, where it explains that the circadian clock the butterflies use sits in their antennae rather than their brains. The correct heading is the antennae one. The magnetism heading belongs to paragraph D, whose first sentence does announce its topic.

The rule: skim to build the map, then close read the paragraph you are actually matching. If two headings survive after the first sentence, the first sentence was never going to decide it, and re-skimming will not help. Read the middle.

#The question type that defeats both

Matching information, the "which paragraph contains the following" type, is the most expensive per mark in IELTS Reading, and it is the one type where neither skimming nor scanning reliably works. From RT021:

"A reference to the role of indigenous knowledge predating scientific discovery."

Neither "indigenous" nor "knowledge" nor "predating" appears in the passage. There is nothing to scan for. The answer is paragraph B, which says local people in the region had long been aware of the butterfly colonies before outsiders located them. Every content word has been replaced.

That is why this type costs so much: you cannot find it, you have to recognise it, which means holding the question in your head while re-reading paragraphs. Treat it accordingly in your budget, and do it last within its passage, when your map of the passage is at its best.

#Which gear, by question type

• True, False, Not Given and Yes, No, Not Given: scan only if there is a proper noun or number. Otherwise skim to the region, then close read three sentences. Never decide from the sentence with the keyword alone.
• Sentence completion, note completion, short answer: scan. These follow passage order, so each answer is below the last one.
• Summary completion from a word box: read the summary first as English and predict the part of speech for each gap before you go near the passage. A gap after "the" needs a noun, and half the box is eliminated before you read anything.
• Summary completion from the passage: scan, but check the word you copy fits grammatically. The most common lost mark here is a correct idea in the wrong form.
• Matching headings: skim for the map, then close read each paragraph. Cross off headings as you use them.
• Matching features: scan the names. Then close read for attribution.
• Matching information: no shortcut. Budget double and do it last.
• Multiple choice with long options: close read the passage region first and answer in your own head, then look at the options. Reading four long options first plants three wrong ideas in your memory.

#The timing arithmetic

Sixty minutes, forty questions, three passages, and in Reading there is no extra transfer time. Listening gives you ten minutes at the end to copy answers onto the sheet. Reading does not. On paper, writing on the answer sheet happens inside the sixty.

Budget it like this:

• Orientation, about 2 minutes per passage: title, first sentences, paragraph letters. 6 minutes total.
• That leaves 54 minutes for 40 questions, an average of 81 seconds each.
• On paper, hold back 2 minutes for transfer, so 52 minutes and 78 seconds a question.

But questions are not equal, and this is the part the standard advice skips. Realistic costs:

• Sentence or note completion, short answer: 45 to 60 seconds
• Summary completion with a word box: 40 to 55 seconds
• Matching features, once located: 45 to 60 seconds
• True, False, Not Given and Yes, No, Not Given: 60 to 75 seconds
• Matching headings: 75 to 90 seconds per paragraph, plus 2 minutes once to read the heading list
• Multiple choice with four long options: 90 to 120 seconds
• Matching information: 90 to 120 seconds

#Why the famous 15, 20, 25 split is often wrong

The usual advice is 15 minutes on Passage 1, 20 on Passage 2, 25 on Passage 3, because difficulty rises with passage number. Cost does not follow difficulty. It follows the question type mix. Price RT021 with the numbers above:

• Passage 1 is 6 True, False, Not Given plus 4 sentence completion plus 3 summary completion. All locate-and-read work. About 15 minutes with orientation.
• Passage 2 is 6 matching headings, 3 matching information, 3 matching features and 1 multiple choice. Twelve of its 13 marks need the whole passage rather than one sentence. About 22 minutes.
• Passage 3 is 5 Yes, No, Not Given, 4 matching information, 3 short answer and 2 multiple choice. About 21 minutes.

Total: about 58 minutes, with two to spare. Notice that the correct split here is roughly 15, 22, 21, not 15, 20, 25. Follow the standard advice on this test and you arrive at Passage 2 with two minutes too few and Passage 3 with four too many.

So spend the first 30 seconds of the test counting question types, not reading. Count the "whole passage" questions in each section: matching headings, matching information, matching features. Those are the expensive marks. Split your working minutes in proportion to where they sit.

#How to recover when you have lost time

Set two checkpoints: minute 20 and minute 40. At minute 20 Passage 1 should be finished and written down. At minute 40 you should be answering Passage 3 questions, not still reading it.

If you are behind, the instinct is to speed up everywhere. That spreads the damage across all forty questions. Cut by type instead, in this order:

1. Abandon the matching information set first. It is the most expensive per mark and, with seven or eight paragraphs, the worst guess in the test. Write a letter in every box and move on. You have just bought five to eight minutes for the price of three or four marks you were unlikely to get.
2. Never abandon completion questions. They are the cheapest marks on the page and they follow passage order, so each one narrows the search for the next.
3. Break passage order in the last 15 minutes. Sweep all three passages for anything with a clean anchor: numbers, names, dates, completion gaps. A 50 second mark in Passage 1 that you skipped is worth more than two more minutes on a Passage 3 multiple choice.
4. Do not re-read a passage you have already read. If you cannot find it the second time, you will not find it the third.

#The last 90 seconds

There is no negative marking, so a blank is strictly worse than a guess. But guesses are not equal, and with 90 seconds left you should fill them in this order:

1. True, False, Not Given and Yes, No, Not Given. One in three.
2. Multiple choice. One in four.
3. Matching headings. One in however many headings are left after you cross off the ones you used.
4. Matching information. One in seven or eight.
5. Completion. Close to zero from a cold guess. Only worth it if you remember a plausible word from the right region, and only in the exact form the sentence needs.

Two heuristics for the forced guesses, both taken from how the traps in our own bank are built. If the statement contains an absolute such as all, only, never, first or alone, FALSE is the better guess, because that is the Quantifier Misdirection pattern. If the statement introduces a detail the passage had no reason to mention, such as someone's motive, funding, family or nationality, NOT GIVEN is the better guess, because that is the Distractor Named Entity pattern. Neither is a rule. Both beat a coin toss.

#How to drill the three gears separately

Doing whole timed tests will show you that you are slow. It will not tell you which gear is broken. Practise them apart.

1. Skim drill. Two minutes on a passage, then close it and write one line per paragraph from memory. Check against the text. You are training the map, not the detail.
2. Scan drill. Take ten completion questions and give yourself 20 seconds each, purely to find the line, not to answer. Speed here is trainable in a week.
3. Close reading drill. Take only True, False, Not Given questions, untimed, and write one sentence for each answer explaining which words in the passage decide it. If you cannot name the words, the answer was a guess even when it was right.
4. The rebuild. Then run a full timed test and check your minute 20 and minute 40 marks.

The free reading practice here runs 355 Academic and General Training reading practices under real exam timing, and every explanation points at the line in the passage that proves the answer, which is what makes drill three possible. The trap-rich passages are the ones to use once you are accurate but slow.

Three more guides worth reading next: the full decision procedure for True, False, Not Given, the paragraph-first method for matching headings, and the minute by minute timing plan.

One last thing. If you are losing marks in Reading, it is worth finding out whether the cause is speed or accuracy before you fix anything, because the two need opposite training. Time yourself on a passage, then redo the questions you got wrong with no clock at all. If they come right, you have a timing problem. If they stay wrong, you have a gear three problem, and going faster will only make it worse.

Ready to put this on the clock? Try a timed reading passage free and see where the minutes actually go.