A research paper is not a story, so stop reading it like one. The finding is at the end, the caveats are in the middle, and the beginning is mostly positioning. Twenty minutes is enough for a working understanding of most papers if you read in the right order.
01Minutes 1 to 3: abstract, then figures
Read the abstract twice. The first read tells you what the authors claim; the second read tells you what they hedge. Then go straight to the figures and read every caption. In most empirical papers the figures are the actual contribution, and everything else is scaffolding around them. If the key plot does not convince you, the prose will not either.
02Minutes 4 to 8: conclusion before methods
Read the conclusion or discussion section next. It states the finding in its strongest defensible form and usually names the limitations the authors could not avoid admitting. Now you know what the paper is trying to prove, which turns the methods section from a wall of procedure into a checklist: does the setup justify that specific claim? Sample size, baselines, and what was held constant are the three things worth checking.
03Minutes 9 to 17: the one section that matters
Every paper has one section that carries the argument, and by now you know which one it is. Read it slowly, mark the load-bearing sentences, and ignore related work entirely on this pass. Related work exists to satisfy the field; it rarely changes what the paper found. If a specific claim depends on a cited result, note the citation and move on rather than chasing it now.
04Minutes 18 to 20: write the proof-of-reading
Close the PDF and write one paragraph: the claim, the strongest piece of evidence, and the biggest catch. Add the one number you would quote in a meeting. This paragraph is the difference between having read the paper and having scrolled it, and it is what you will wish you had written when the paper comes up three weeks later.
What if the paper is outside my field?
Spend the first five minutes on the abstract and figure captions only, then find a survey or the paper's own introduction for vocabulary. Out-of-field papers take two sittings, not one longer one.
Should I read the math?
On the first pass, no. Verify the setup and the claim first. Work through derivations only when you have decided the paper matters to your work, because that is an hours-level commitment.
How do I keep track of papers I have read?
Keep the one-paragraph summaries in a single running document with the paper title and a link. Searchable proof-of-reading beats a folder of annotated PDFs you will never reopen.
Summarizing at volume, not one page at a time?
Summarize Pro batches PDFs, papers and reports, every key claim cited to its source page, exports to Word or Excel. The free tool is for one page; Pro is for the pile.