Skimming and scanning: two different skills, two different jobs
These two words are used interchangeably and they should not be. They are different operations with different eye movements, and picking the wrong one is why “I skimmed it” so often means “I did not find what I needed”.
Skimming is reading for the gist: you cover the whole text at low resolution to find out what it says. Scanning is searching: you ignore the meaning entirely and hunt for one specific thing.
Knowing which one you are doing before you start is most of the technique.
How to skim properly
Skimming is not reading fast. It is reading selectively, which means deciding in advance which parts you will read at full speed.
Read the title, the introduction in full, then the first sentence of each paragraph, then the conclusion in full. Add any heading, bolded term, figure caption and numbered list along the way. On most expository writing this is around twenty per cent of the words and gets you most of the argument.
The first-sentence rule works because of how paragraphs are constructed — but the main idea is often not the first sentence, so when a paragraph opens with a transition or a question, read one more line before moving on.
Stop skimming when you hit something you need. Skimming is a mode you enter and leave, not a speed you set for the whole document.
How to scan properly
Scanning is a visual search task, and the technique is the opposite of reading: you actively suppress comprehension so it does not slow you down.
Decide the exact target first — a date, a name, a figure, a term. Then hold its shape in mind rather than its meaning. Numbers have a distinctive shape among words. Proper nouns carry capitals. A term you are hunting has a length and a silhouette.
Move your eyes down the middle of the column in a zigzag rather than left to right along each line. You are not reading the lines; you are waiting for the shape to pop out. If you find yourself understanding sentences, you have dropped back into reading and lost the speed advantage.
Use the structure: indexes, headings, bold terms and the fact that most documents put definitions near their first use.
Choosing between them
The question to ask is whether you already know what you are looking for:
Skim when you do not know what is in there
Deciding whether a source is relevant, previewing a chapter before reading it properly, getting the shape of an argument before a seminar.
Scan when you know exactly what you want
Finding the figure for a citation, locating where a term is defined, checking whether a paper mentions a particular method.
Do neither when accuracy matters
Anything you will be assessed on, quote, or act on. Both techniques trade accuracy for speed, and both will confidently miss a qualifying clause.
Triage: using both on a reading list
The practical case is a stack of sources and not enough hours. The sequence that works is scan, skim, read — cheapest filter first.
Scan each source for the terms central to your question. Anything that does not contain them at all is out, and that is usually a third of the list for a few minutes of work.
Skim what survives, to find out what each one actually argues. Now you can rank them, and you will typically find three or four that matter and a long tail that is background.
Read those three or four properly. For papers specifically, the three-pass method is a more structured version of exactly this.
Where this breaks down
Skimming assumes conventional structure. It works on textbooks, reports, journalism and papers because those put topic sentences first and signpost their sections. It fails badly on fiction, on argumentative essays that withhold their thesis deliberately, and on anything where the point is cumulative.
It also fails when you do not know the vocabulary. Skimming works by recognising which words are load-bearing, and in an unfamiliar field every word looks equally important — so a skim of a paper in a field you do not know produces a confident and entirely wrong impression.
Automatic highlighting is a reasonable substitute in that situation, because statistical ranking does not care whether you recognise the terms: it finds the sentences that are central to that document and marks the figures and definitions. It gives you a non-expert triage pass that your own skimming cannot.