Research metrics, clearly
Why Google Scholar, Scopus, and Web of Science Citation Counts Differ
The same researcher can have three different citation totals. Learn what each database covers, why the numbers move differently, and how to read them responsibly.
Seeing different citation totals for the same researcher is normal. Google Scholar, Scopus, Web of Science, ResearchGate, and OpenAlex do not observe one universal citation ledger. Each service builds its own collection of works, matches authors and references in its own way, and recalculates metrics on its own schedule.
The useful question is not “Which number is the real one?” It is “What evidence does this number describe?”
Coverage changes the result
Google Scholar searches broadly across scholarly material on the web. Its profiles calculate total citations, h-index, and i10-index as Scholar finds citations to works associated with an author. That broad discovery can include journal articles alongside theses, books, preprints, reports, and repository copies.
Scopus builds author profiles from publications indexed in Scopus. Those profiles include publications, citations, affiliations, subject areas, and co-authors. A document that is not represented in Scopus cannot contribute to the same profile in the same way.
Web of Science states that aggregate profile metrics are derived from citations to Web of Science Core Collection indexed papers. Its h-index is calculated from Core Collection publications and citations attached to the profile. Clarivate also notes that this can produce a lower count than a broader source such as Google Scholar.
OpenAlex constructs citation links by extracting reference lists and matching them to existing works, first by DOI when possible and then by other bibliographic metadata. Its author records expose works, citations, h-index, and i10-index, but those figures reflect the OpenAlex graph and its matching process.
Different coverage therefore produces different denominators, different citing works, and different totals.
Author matching also matters
Names are ambiguous. Two researchers can share a name, one researcher can publish under several forms of a name, and affiliations change. A service may split one person into two profiles or attach a similarly named author’s paper to the wrong profile.
Before interpreting a difference, inspect the publication lists. Look for:
- missing papers;
- duplicate versions of the same paper;
- papers belonging to a different author;
- recent publications that have not been indexed yet;
- merged or corrected records.
Google Scholar specifically recommends reviewing suggested articles and merging duplicates. Scopus provides profile-correction workflows when documents or citations are missing. A corrected publication list can change both the citation total and derivative metrics such as h-index.
Updates do not arrive together
A citing paper must be discovered, parsed, matched to the cited work, and attached to the correct author. Each source performs those steps independently. A new citation can therefore appear in one service days or weeks before another, or never appear in a source whose collection does not include the citing document.
Counts can also decrease. Duplicate records may be merged, incorrect papers removed, or suspicious and unmatched citations reprocessed. A decrease is not automatically evidence that a researcher lost legitimate influence.
How to compare metrics responsibly
Treat each source as a separate observation:
- Record the provider and date with every number.
- Compare a provider with its own earlier reading, not with a different provider’s total.
- Inspect the underlying publications when a movement looks unusual.
- Use the source expected by the decision you are supporting, such as a specific institutional assessment.
- Never add the citation totals from several providers together. Their coverage overlaps, so the result would double-count many citations.
A Google Scholar citation count and a Web of Science citation count can both be correct descriptions of different collections.
Citatum follows this principle by keeping every metric attached to its provider. It can help you revisit source-specific readings, but it does not convert unlike datasets into a single score.
The short answer
Citation counts differ because the databases index different material, resolve authors and references differently, and update on different schedules. Keep the source and observation date beside the number. That context is part of the metric, not an optional footnote.