Home › Blog › Bibliometric analysis
What is bibliometric analysis? A beginner's guide for 2026
Bibliometric analysis is a method that uses the statistics of publications - citations, keywords, authors and co-authorship - to map how a research field has developed and where it is heading. In short: instead of reading 800 papers, you analyse the patterns across all of them.
When should you use bibliometric analysis?
Use it when your research question is about a field rather than a single phenomenon - for example "how has research on greenwashing evolved over 25 years?" or "what are the emerging themes in B2B AI marketing?". It is ideal for a first publishable paper because the data is structured, the method is transparent, and the workflow is repeatable. It is not the right tool when you need to test a hypothesis about behaviour - that calls for a survey and SEM, or an experiment.
The two main analyses
Bibliometric studies usually combine two layers. Performance analysis describes the field: most-cited papers, most productive authors, leading journals and countries. Science mapping reveals structure and relationships: co-citation analysis, bibliographic coupling, co-authorship networks and keyword co-occurrence. Most strong papers report both, then interpret what the maps mean for future research.
VOSviewer vs Biblioshiny: which should a beginner use?
Both are free. VOSviewer is point-and-click and produces the cluster network maps you see in published papers - the fastest way to get a publishable visual. Biblioshiny (the web interface for the Bibliometrix R package) does more of the performance analysis and thematic evolution out of the box, and exports cleaner tables. A practical answer: start with VOSviewer for the maps, add Biblioshiny for the descriptive tables and three-field plots. No coding is required for either if you use Biblioshiny rather than raw R.
The workflow, step by step
- Define scope and search string. Pick a database (Scopus or Web of Science) and write a precise query - keywords, fields, year range.
- Apply PRISMA-style filtering. Document how many records you found, screened, and kept, with clear inclusion and exclusion criteria. Reviewers expect this flow diagram.
- Export the dataset. Download full records and citations in the format your tool needs (CSV or plain text).
- Run performance analysis. Top papers, authors, journals, countries, annual output.
- Run science mapping. Co-citation, coupling, co-word and co-authorship networks.
- Interpret and write. The clusters are not the finding - your interpretation of them is. Translate each cluster into a research theme and propose a future agenda.
How a bibliometric paper is usually structured
A common, reviewer-friendly structure is: Introduction and research questions; Methodology (database, search string, PRISMA flow, tools); Performance analysis results; Science mapping results; Discussion organised around the identified themes; and a future research agenda. A widely used framing for the whole paper is the TCCM lens - Theory, Context, Characteristics and Methodology - which gives the discussion a clear shape.
The single most common reason bibliometric papers get rejected is that they stop at description - pretty maps, no insight. The contribution is your interpretation and the agenda you set, not the network diagram.
Learn this hands-on
If you'd rather be walked through it on a real dataset, the free Research Methods Workshop seminar often covers bibliometric papers, and the Bibliometric Analysis & Paper Writing workshop takes you from search string to submission-ready manuscript. No coding background needed.
