Fit problems
The paper is sound but wrong for the journal — out of scope, wrong methodology for that readership, or aimed at an audience the journal does not serve. Nothing about the research needs to change; the target does.
Fit problems are the cheapest to avoid and the most commonly ignored. Reading three or four recent issues of a target journal before submitting tells you more about fit than any scope statement on the website.
A fit rejection says nothing about quality. The same manuscript may be accepted unchanged at a better-matched journal.
Contribution problems
The reviewers understood the work and did not think it added enough. This may mean the increment over existing literature is genuinely small, or that the manuscript failed to articulate an increment that is actually there.
The second case is far more common than authors expect. Work that reads as incremental is often work whose framing buried the interesting part — a result presented as a minor extension when it in fact resolves something the field had left open.
Distinguishing the two matters. If the contribution is real but badly framed, that is a rewriting job. If it is genuinely thin, more writing will not fix it, and the honest answer is more research or a more appropriate venue.
Method and evidence problems
Reviewers identified something they consider a flaw: a design that cannot support the claim, missing controls, statistical treatment that does not match the data, or conclusions reaching past what the results show.
Some of these are fatal for the paper as written. Many are not — reviewers frequently ask for analyses the authors can run, or for claims to be scaled back to what the evidence supports. A major-revision decision dressed as a rejection is worth reading carefully before you give up on the venue.
Presentation and process problems
Language that obscures the argument, a structure that hides the findings, figures that cannot be read at print size, references that do not match the style. Individually minor, collectively enough to lose an editor before the science is assessed.
And the avoidable process failures: ignoring word limits, omitting an ethics statement where one is required, missing a data availability statement, or submitting without the declarations the journal asks for. These do not reflect on the research at all, and they are entirely within your control.