How to Do a Literature Search: A Step-by-Step Guide for Researchers

How do you search the research literature properly?

Typing a few words into Google Scholar or a research database can find papers. That does not necessarily mean you have conducted a good literature search.

A useful literature search has a clear purpose. It starts with a question, translates that question into searchable concepts, looks in appropriate information sources, uses search terms deliberately, and records enough of the process that you can understand what you did later.

The challenge is finding the right balance. Search too broadly and you may face thousands of loosely relevant records. Search too narrowly and you may miss studies that use different terminology, older terminology, alternative spellings, or different indexing terms.

The goal is not simply to retrieve as many papers as possible. It is to search systematically enough for the purpose of your project while keeping the results relevant and manageable.

A good literature search is built, tested, and refined

A literature search is therefore better treated as an iterative process than as a single query. Your first search teaches you something about the terminology of the field, and that information helps you improve the next search.

How to conduct a literature search step by step

Step 1: Decide what the search needs to accomplish

Before choosing keywords or databases, clarify why you are searching.

A preliminary search for a research proposal does not necessarily require the same level of comprehensiveness and documentation as a search supporting a systematic review. You might be searching to understand a new topic, identify a research gap, develop a methodology, find previous studies using a particular measure, prepare a literature review, or identify all eligible studies for an evidence synthesis.

The purpose determines how comprehensive, reproducible, and carefully documented your search needs to be.

Step 2: Turn your topic into a focused question

Broad topics are difficult to search efficiently. A topic such as “social media and students” can produce an enormous and poorly defined literature.

A more focused question might ask whether social media use is associated with academic performance among university students. That question immediately reveals concepts that can be translated into search terms.

Question frameworks can help when they fit the research problem. PICO, for example, is widely used for questions involving a population, intervention, comparator, and outcome. Other disciplines and question types use different frameworks. You do not need to force every research question into PICO simply because it is familiar.

Step 3: Identify the main searchable concepts

Break the question into its core concepts rather than searching the entire question as a sentence.

For the example question about social media and academic performance among university students, possible concepts are:

• social media;
• university students; and
• academic performance.

Not every element of a research question necessarily belongs in the database search. Adding too many concepts can make a search unnecessarily restrictive because a relevant paper may not mention or index every concept in the way you expect. Guidance for evidence synthesis likewise emphasizes building searches around the principal concepts rather than automatically searching every element of a question.

Step 4: Build a term bank for each concept

Researchers and authors do not always use the same words for the same idea. A strong search therefore usually needs more than one term for each concept.

For each concept, consider:

• synonyms;
• alternative spellings;
• abbreviations and acronyms;
• older and newer terminology;
• singular and plural forms where relevant;
• closely related expressions; and
• database-specific controlled vocabulary.

Your term bank should develop as you search. Relevant papers can reveal terminology you did not initially consider.

Step 5: Look for controlled vocabulary as well as keywords

Many bibliographic databases assign standardized subject terms to records. In MEDLINE, for example, the National Library of Medicine uses Medical Subject Headings, or MeSH, a hierarchical controlled vocabulary for indexing biomedical literature.

Keyword searching and controlled-vocabulary searching solve different problems. Keywords help retrieve the language authors actually use in titles and abstracts, including terminology that may be new or not yet represented adequately in an indexing system. Controlled vocabulary can bring together records about the same concept even when authors use different words.

For a thorough database search, it is often useful to investigate both.

Step 6: Choose databases that match the subject

No single database should automatically be treated as the complete scholarly literature. Databases differ in subject coverage, publication types, journals, geographic representation, indexing practices, and search functionality.

Choose databases because they cover the literature relevant to your question, not simply because they are familiar or easy to access.

Depending on the field, researchers may use multidisciplinary citation databases, discipline-specific bibliographic databases, institutional databases, or specialized collections. A biomedical project, for example, may require different sources from a project in engineering, education, economics, or the humanities.

The number of databases you need also depends on the purpose of the search. A preliminary exploration may begin with one major database. A systematic evidence synthesis may require multiple databases and additional search methods.

Step 7: Combine synonyms with OR

Boolean operators tell a database how different terms relate to one another.

Use OR to combine alternative terms representing the same concept:

("social media" OR "social networking" OR "social networking sites")

OR broadens the search because a record can match any of the alternatives.

Step 8: Combine different concepts with AND

Use AND to connect separate concepts:

("social media" OR "social networking") AND ("university students" OR "college students")

AND narrows the search because retrieved records must satisfy both concept groups. This basic pattern, OR within concepts and AND between concepts, is a standard way to construct multi-concept searches.

Parentheses matter because they tell the database which terms should be treated as a group. Database syntax varies, so always check the search help for the platform you are using.

Step 9: Use NOT cautiously

NOT excludes records containing a term. That can look useful when irrelevant results dominate a search, but it can also remove relevant papers that happen to mention the excluded word.

Step 10: Learn the syntax of the database you are using

Search strategies are not always portable by simple copy and paste. Databases and search platforms can differ in field codes, phrase searching, truncation symbols, wildcard rules, proximity operators, controlled vocabulary, and other syntax.

For example, PubMed supports Boolean operators, field searching, phrase searching, wildcards, filters, and other search functions. Other databases may express equivalent functions differently.

When moving a strategy from one database to another, translate it rather than assuming the original syntax will work unchanged.

Step 11: Run a pilot search and inspect what you retrieve

Your first search is a test, not a final answer.

Open several highly relevant records and inspect their titles, abstracts, keywords, subject headings, and terminology. Ask:

• Are known relevant papers appearing?
• Are important synonyms missing?
• Are unexpected terms repeatedly appearing in relevant papers?
• Is one concept making the search too restrictive?
• Are irrelevant records revealing an ambiguity in one of your terms?

This is where searching becomes iterative. Add useful terminology, remove misleading terms, adjust concept groups, and test again.

Step 12: Refine for sensitivity and precision

Search quality involves a trade-off between finding relevant material and limiting irrelevant material.

If you retrieve too few results, consider adding synonyms, checking spelling and syntax, removing an unnecessarily restrictive concept, broadening field restrictions, or identifying additional controlled vocabulary.

If you retrieve far too many irrelevant results, inspect why. You may need a more specific term, an additional essential concept, a field restriction, a more precise subject heading, or another database-specific technique.

A large number of results is not automatically evidence of a bad search, and a small number is not automatically evidence of a good one. Relevance to the question matters more than an arbitrary target number.

Step 13: Search beyond the first database query when appropriate

Database searching may be only one route to relevant literature. Depending on the project, useful supplementary methods can include examining the reference lists of relevant papers, looking at papers that cite an important study, searching for key authors, and investigating appropriate sources of grey literature.

These approaches can reveal records that a keyword search missed because of terminology, indexing, or database coverage.

Step 14: Export, organize, and deduplicate your records

Once the search starts producing useful results, avoid managing everything manually in browser tabs.

Export records to a reference manager or other organized system. Preserve bibliographic information and, where available, identifiers such as DOIs or database accession numbers. When searching multiple databases, identify and remove duplicate records before screening so that the same publication is not repeatedly assessed.

Step 15: Document the search while you are doing it

Do not rely on memory to reconstruct a complex search weeks later.

At minimum, record:

• the database and search platform;
• the date searched;
• the complete search query or strategy;
• important limits or filters;
• the number of results retrieved; and
• any major changes made during refinement.

For systematic reviews and other evidence syntheses, reporting requirements are substantially more rigorous. PRISMA-S was developed specifically to improve transparent and reproducible reporting of literature searches used in systematic reviews.

For high-stakes evidence syntheses, consider involving an information specialist or librarian. Formal search strategies can also be peer reviewed; the PRESS guideline provides a structured approach for reviewing elements such as translation of the question, Boolean logic, subject headings, free-text terms, spelling and syntax.

From a research question to a working search strategy

This is a starting strategy, not a universal search string. A real search would need to be adapted to the database, the exact research question, and the level of comprehensiveness required.

Common literature-search mistakes that can distort what you find

Match the search process to the purpose of your research

You do not need the same search procedure for every task. What matters is making the search rigorous enough for the claim you intend to make about the literature.

The most transferable habit is simple: preserve your search strategy. A saved search becomes something you can inspect, improve, rerun, explain, and, when appropriate, reproduce.

Literature search checklist

Frequently asked questions about literature searching

A literature search is a research process, not a single search box

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.