What Is Snowball Searching, and When Should You Use It?

What Does It Mean to “Snowball” a Literature Search?

You find one paper that is almost exactly about your topic. Its reference list contains several promising studies. You follow those references, find more relevant papers, inspect their references and citations, and suddenly your search has expanded far beyond the original article.

Researchers often describe this process as snowball searching or simply snowballing. The metaphor is intuitive: you begin with a relatively small set of relevant papers, then progressively identify additional literature through connections among them.

What can be confusing is that “snowballing” is not used with exactly the same meaning everywhere. In some research traditions it refers specifically to repeated backward and forward citation searching. Elsewhere it is used more loosely for citation-based literature discovery. Understanding the underlying procedure matters more than relying on the label alone.

Snowball Searching Builds Outward From Relevant Papers

How Snowball Searching Actually Works

You Begin With One or More Relevant Seed Papers

Snowball searching does not usually begin with an empty search box. It begins with literature you already know is relevant.

These starting publications are often called seed papers or, in more standardized citation-searching terminology, seed references. They might come from a database search, a review article, a paper recommended by a knowledgeable colleague, or previous familiarity with the field.

The quality of these starting points matters because citation searching follows relationships surrounding them. A highly relevant seed paper can lead into a productive literature network. A paper that only touches your topic tangentially may instead produce many weakly relevant leads.

The First Direction Is Backward

One way to expand from a seed paper is to examine what it cited. This is backward citation searching.

Suppose your seed paper was published in 2022. Its reference list might lead you to studies from 2020, 2017, 2011, or considerably earlier. Some may use different terminology, come from neighboring disciplines, or represent theoretical and methodological foundations that did not appear in your original keyword results.

Backward searching can therefore help reconstruct how a research problem developed. It is particularly useful when you suspect that important earlier work exists but cannot easily retrieve it using contemporary terminology.

The Second Direction Is Forward

You can also ask which later publications cited your seed paper. This is forward citation searching.

Forward searching requires a citation source capable of identifying citing documents. Citation indexes and scholarly search services such as Scopus, Web of Science, and Google Scholar can support this process, although their coverage differs.

These later publications may extend the original study, apply its ideas in another context, replicate it, criticize it, use its methods, or cite it for a much smaller reason. A citation is therefore a discovery lead, not evidence that the citing paper is relevant.

The “Snowball” Comes From Repeating the Process

Finding papers through one round of citation searching is only the beginning of what many researchers mean by snowballing.

You screen the newly discovered papers. When one is genuinely relevant, it can become another seed. You then examine its references, its citing publications, or both. Newly identified relevant papers can in turn become starting points for another round.

This iterative expansion distinguishes snowball searching from casually looking at a few references while reading. In current systematic-search terminology, iterative citation searching is a clearer description for one or more repetitions of citation-based searching. The TARCiS statement recommends that terminology because words such as “snowballing” have been used inconsistently across the literature.

Why the Terminology Can Be Confusing

Researchers do not always use “snowballing,” “citation chaining,” “citation chasing,” “citation tracking,” and related expressions consistently. A methodological scoping review found substantial variation in terminology, and the subsequent TARCiS guidance recommended more standardized terms for reporting citation-searching methods.

That does not make “snowball searching” an unusable expression. It remains readily understandable and is established in parts of the literature. But when describing a search method in a thesis, review, or publication, saying exactly what you did is more informative than writing only “we used snowballing.”

For example, reporting that you conducted backward and forward citation searches from specified seed references and repeated the process for newly eligible papers conveys much more methodological information.

Snowball Searching Can Find What Keywords Miss

Keyword searching depends heavily on language. Your search strategy has to anticipate the terminology authors used in searchable bibliographic fields.

That becomes difficult when a concept has multiple names, terminology has changed historically, disciplinary vocabularies differ, or the idea you care about is not prominent enough to appear in the title or abstract.

Citation relationships offer another route. Two papers can be intellectually connected even if their authors describe the topic differently. This is one reason citation-based searching can uncover research missed by keyword searches.

The TARCiS statement specifically recommends seriously considering backward and forward citation searching as supplementary techniques for systematic-search topics that are difficult to search using text-based methods.

It Can Be Particularly Helpful When Terminology Changes Over Time

Imagine searching a contemporary concept whose present name became common only recently. Earlier researchers may have investigated something very similar under terminology that would never occur to you when constructing your first search string.

A relevant modern paper may cite that older literature even when the vocabulary differs. Following its references can therefore help you find important older research written before modern terminology existed.

The same principle applies across disciplines. Researchers may investigate closely related phenomena while using different conceptual traditions and labels. Citation links can sometimes expose these connections more effectively than repeatedly adding synonyms to an increasingly unwieldy search string. Anyone who has built a Boolean query long enough to require its own table of contents may appreciate the attraction.

Snowball Searching Can Help Reveal the Structure of a Field

Snowball searching does more than increase the number of records you have found. Following citation relationships can also help you understand how a body of scholarship is connected.

You may notice particular papers recurring across reference lists. Older studies may appear to anchor a theoretical tradition. Later publications may extend a particular method. Separate clusters may reveal different disciplinary approaches to what initially seemed like one research problem.

These patterns can provide leads when you are trying to identify potentially foundational or influential studies. They should still be interpreted carefully. Repeated citation is evidence of scholarly attention or connection, not automatic proof that a paper is seminal, correct, or methodologically superior.

Snowball Searching Is Not the Same as Searching for Similar Papers

Some databases and discovery systems provide “related articles,” “similar papers,” or algorithmic recommendations. Those features can be useful, but they are not necessarily based on direct citation relationships.

A similarity system might compare textual content, keywords, citation patterns, embeddings, readership behavior, or other signals depending on the platform. Snowball searching, by contrast, is conventionally concerned with following identifiable relationships among cited and citing publications.

This distinction matters because related-article and similar-paper searches may retrieve useful literature through a different mechanism.

Snowball Searching Does Not Automatically Produce an Unbiased Sample of Literature

A citation network is not a neutral map of everything ever written about a topic. Authors choose what to cite, databases differ in what they index, fields have different citation practices, and older or more visible publications have had different opportunities to accumulate citations.

There is also a starting-point problem. If your seeds all represent one theoretical school, geographical region, discipline, language, or group of researchers, repeatedly following their citation networks may keep returning you to the same scholarly neighborhood.

It Is Usually a Complement, Not an Automatic Replacement for Database Searching

For many research purposes, snowball searching works best alongside other discovery methods rather than replacing them without justification.

A database search can identify records based on predefined concepts across a broad indexed collection. Citation searching then provides a different retrieval pathway based on relationships among publications. The two methods therefore have different strengths and failure points.

This distinction becomes especially important in systematic reviews and other evidence syntheses. Citation searching may identify unique eligible records that database searches missed, but the effectiveness varies by topic. Search methods should therefore be chosen according to the review question and methodological requirements rather than assuming that one technique is universally superior.

How One Paper Can Grow Into a Literature Network

The important feature is not that every connected publication was retained. The search expanded through citation relationships, while the research question and eligibility criteria continued to determine what belonged in the evidence base.

Common Misunderstandings About Snowball Searching

When Is Snowball Searching Worth Using?

Snowball searching is most valuable when citation relationships can answer a discovery problem that keyword searching does not handle well. The stronger the seed papers and the more difficult the topic is to express consistently in searchable terminology, the more useful citation-based exploration may become.

Be More Systematic When the Search Supports a Review

There is an important difference between using snowball searching to become familiar with a field and using it as part of the search methodology for an evidence synthesis.

For exploratory reading, you can follow promising citations pragmatically. For a systematic or scoping review, decisions about seeds, search directions, tools, iterations, eligibility, and stopping should be sufficiently explicit that readers can understand how additional studies were sought.

Current TARCiS guidance provides recommendations for the terminology, application, and reporting of citation searching. Where citation searching forms part of a review methodology, consulting such guidance is preferable to treating snowballing as a self-explanatory procedure.

Know When the Search Is Drifting

One of the more interesting outcomes of iterative citation searching is discovering that a relevant paper connects to a literature you did not know existed. Sometimes this is exactly the breakthrough you needed. Sometimes it is scope creep wearing an unusually convincing reference list.

If one paper leads into an apparently different literature, return to your research question. Ask whether the new branch addresses the same phenomenon, provides a necessary conceptual foundation, or represents an interesting but ultimately different problem.

Decide How You Will Stop

Iterative searching can theoretically continue through many generations of citations. In practice, time and relevance impose limits.

Later rounds may yield mostly duplicates or records that fail your eligibility criteria. At that point, the marginal value of another iteration may become small. The appropriate stopping rule depends on the purpose and rigor of the search, so it is better to establish a defensible approach than to stop merely because the reference manager has become frightening.

For substantial searches, think explicitly about how far citation chaining should continue before additional rounds become unproductive.

Before You Start Snowball Searching, Check These Points

Questions Researchers Ask About Snowball Searching

Snowball Searching Follows the Literature Through Its Connections

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.