16 September 2026 · 7 min read
An assignment with twelve sources can be referenced by hand. A thesis with eighty, written over six months, restructured twice and cut for length in the final week, cannot. What breaks is never the first draft — it is the fourth, when a paragraph moves, two sources get dropped, and the reference list still lists them.
Zotero and Mendeley are both free, both work offline once installed, and both do the same three useful things: store the source and its details, insert the in-text citation as you write, and rebuild the reference list from whatever is actually cited in the document. That third one is the point. A generated list cannot contain a source you deleted, and cannot omit one you added at midnight.
Pick the exact style your department requires — APA 7th, Harvard as your college defines it, IEEE, Vancouver — and set it in the plugin before you write. Check the output against one worked example from your department's guideline, because “Harvard” is not one style and Nepali colleges each document their own variant.
Then fix the metadata as you save each source. This is the step everyone skips and everyone pays for: an automatically captured record often has the journal name in title case, an editor in the author field, or “n.d.” where the year should be. Two minutes at capture time beats an evening of correcting a generated list that is wrong in eighty places identically.
Nepali theses lean on material that reference managers capture badly, and these are the entries examiners notice because they look wrong at a glance.
Whatever the manager produces, you are responsible for what appears in the document. A generated reference list is a first draft, not a finished one.
The most serious citation problem in a thesis is not formatting. It is citing a study you have only seen quoted in someone else's literature review. It is quick, it is common, and it is how errors propagate — a misreported finding gets copied down a chain of theses for a decade. If you cannot open it, either find it or cite it as cited in the source you did read, in the form your style specifies.
Examiners test this at viva by picking a reference and asking what it argued. It is an easy question if you read the paper and an unanswerable one if you did not, and there is no recovering the room afterwards.
Refresh the bibliography, then read it against the document. Every in-text citation must appear in the list and every entry in the list must be cited somewhere — a source in the list that appears nowhere in the text is the classic symptom of a chapter that got cut. Check that years match between text and list, that author spellings are consistent, and that no “(Author, n.d.)” survived.
Then run the similarity check and look specifically at the reference list. It will match heavily, because a correct bibliography matches every other document citing the same works. That is the report doing its job, and many departments exclude the list from the score — but it is worth being able to say so calmly rather than discovering it as a shock.
Related:
If your citations and your reference list have already drifted apart, send the document — reconciling them is a specific, finishable job and it is worth doing before your supervisor reads the chapter.