03 · What You Need to Know
The Three Terms Answer Different Questions About Search Quality
The distinctions become easier if you ask what each term is actually evaluating.
Systematic: Was the evidence identified through a planned, methodical process appropriate to a defined search objective?
Reproducible: Could another researcher understand what was done and repeat the search as closely as the search system permits?
Comprehensive: Was the search broad and sensitive enough, across appropriate sources and methods, to capture the relevant evidence required by the research purpose?
These questions overlap, but they are not substitutes for one another.
| Quality |
Central concern |
What it does not guarantee |
| Systematic |
A planned, structured, transparent approach to identifying a defined evidence base |
That every relevant study was found |
| Reproducible |
Sufficient documentation for the search to be repeated as closely as possible |
That the search was conceptually sound or sufficiently broad |
| Comprehensive |
Extensive coverage of the relevant evidence space |
That the process was systematic or reproducibly reported |
What Makes a Search Systematic?
Systematic searching is more specific than searching carefully or being organized.
Gusenbauer and Haddaway distinguish systematic searching from lookup and exploratory searching. In their formulation, systematic searching aims to identify the relevant records on a defined topic within resource constraints through transparent and reproducible procedures. The information need is already sufficiently developed, and the search proceeds through planned search steps rather than open-ended exploration.
This does not mean exploration disappears from systematic-review work. Quite the opposite: exploratory searching can be necessary before the formal systematic search is finalized because researchers need to understand terminology, concepts, and the likely boundaries of the evidence base.
The distinction is temporal and methodological. You may explore first to learn how to search. The eventual systematic phase then applies a defined method to identify the evidence required by the review.
This is why not every literature search needs to be systematic. Systematic searching is appropriate when the research requires this kind of planned evidence identification, not simply whenever researchers want to be careful.
Systematic Searching Is About the Process, Not the Appearance of the Query
A long Boolean string does not make a search systematic.
You can construct a complicated query with dozens of synonyms, controlled vocabulary terms, proximity operators, and nested parentheses while still searching an inappropriate database or omitting an important concept. Conversely, a relatively simple search may be entirely appropriate when the phenomenon has distinctive terminology.
The systematic character of the search comes from the relationship among the research question, eligibility criteria, concepts, evidence sources, search methods, testing, documentation, and reporting.
Cochrane, for example, describes systematic reviews as requiring a thorough, objective, and reproducible search across a range of sources to identify as many eligible studies as possible within resource constraints. Database selection should be guided by the review topic, and the search should be sufficiently extensive to reduce the risk that the retrieved evidence is an unrepresentative subset of what exists.
A search strategy therefore involves substantially more than the query entered into a search box.
What Makes a Search Reproducible?
Reproducibility concerns whether the search can be reconstructed and repeated.
Imagine reading this methods statement:
“Relevant studies were identified through searches of PubMed, Scopus, and Google Scholar.”
You know where the authors searched, but not enough to repeat what they did.
You would need additional information such as the search date, the actual queries or strategies, database platforms where relevant, limits and filters, and potentially details of supplementary methods such as citation searching or website searching.
Cochrane requires the search process to be documented in enough detail that database searches are reproducible to the extent possible. It distinguishes internal documentation from reporting in the eventual review, while emphasizing that adequate record-keeping is necessary for correct reporting.
PRISMA-S similarly provides reporting guidance for literature searches in systematic reviews, covering information sources, search strategies, limits, supplementary searches, dates, and other information needed to understand how evidence was identified.
Reproducible Does Not Mean Another Researcher Will Always Get Identical Results
The word reproducible can sound stronger than search technology allows.
A bibliographic database queried using the same strategy may allow a close reconstruction of the original search, although its contents can change as records are added, removed, corrected, or reindexed.
Web search engines create additional difficulties. Search algorithms change. Results may depend on location, personalization, indexing changes, and other system characteristics. Cochrane consequently notes that web searches are not reproducible to the same extent as bibliographic database searches.
Reproducibility in literature searching should therefore be interpreted practically. The objective is to preserve enough information that another researcher can understand and repeat the procedure as closely as the source permits.
Watch Out
Do not promise exact reproducibility when the search system itself cannot provide stable results. Report the platform, query, date, limits, and procedures accurately, and distinguish reproducible search methods from identical future retrieval.
A Reproducible Search Can Still Be a Bad Search
This distinction is particularly important.
Suppose a researcher documents the following search perfectly:
Database: one database with poor coverage of the relevant discipline.
Query: an exact phrase that captures only one way of describing the phenomenon.
Date: precisely recorded.
Limits: clearly stated.
Another researcher could reproduce it beautifully.
They would reproduce the same methodological weakness.
Reproducibility tells you that the procedure is inspectable and repeatable. It does not establish that the search concepts were valid, that the sources were appropriate, that important synonyms were included, or that the relevant evidence was adequately covered.
This is why transparency is necessary for evaluating search quality but cannot substitute for search quality.
What Makes a Search Comprehensive?
Comprehensiveness concerns coverage.
A comprehensive search attempts to search sufficiently broadly and sensitively that the relevant evidence required by the research purpose is unlikely to be systematically missed.
That may involve searching multiple appropriate bibliographic databases, using numerous legitimate terms for important concepts, combining controlled vocabulary and free-text terms, searching trial registers or other specialized sources, following citations, examining references, searching grey literature, or using additional methods appropriate to the topic and review methodology.
Cochrane describes its reviews as taking a systematic and comprehensive approach to identifying studies meeting eligibility criteria. It recommends searches that are as extensive as possible to reduce reporting bias and identify as much relevant evidence as possible.
The phrase “as extensive as possible” still operates within methodological and practical boundaries. Comprehensiveness is not equivalent to searching every database that exists.
Comprehensive Does Not Mean Search Everywhere
A search becomes comprehensive by covering the relevant evidence space, not by accumulating sources indiscriminately.
Suppose a review concerns a specialized educational intervention. Searching ten biomedical databases that contain little relevant education research would not necessarily make the search more comprehensive than searching a smaller set of well-chosen education, psychology, multidisciplinary, and supplementary sources.
Database selection should follow the question and the distribution of relevant evidence.
Likewise, adding every imaginable synonym can decrease conceptual precision without meaningfully improving coverage. A term that sounds related may represent a different construct and introduce large quantities of irrelevant literature.
Comprehensiveness therefore requires informed boundaries. You need to know what evidence should be covered before you can judge whether the search covers it adequately.
Comprehensiveness Is a Goal, Not a Demonstrable State of Perfection
There is a fundamental difficulty in claiming that a search found everything relevant: if you already knew every relevant study that existed, you would not need the search to discover them.
Perfect recall is therefore difficult to demonstrate against an unknown evidence universe.
Relevant studies can remain undiscovered because of inconsistent terminology, inadequate indexing, database coverage, inaccessible sources, publication practices, language restrictions, missing metadata, or other retrieval problems.
This is why rigorous guidance usually frames comprehensive searching as an attempt to identify as much relevant evidence as possible rather than as a guarantee of total retrieval.
Searchers can strengthen confidence by using multiple complementary sources and methods, testing strategies against known relevant studies, examining whether important terminology is represented, using citation searching, and involving information specialists where appropriate.
But “comprehensive” should remain a methodological claim about the scope and intent of the search, not a declaration that no undiscovered relevant study could exist.
Sensitivity Helps Explain What Comprehensive Retrieval Is Trying to Achieve
In information retrieval, sensitivity, also called recall, describes the proportion of relevant records in a source that a search successfully retrieves.
Cochrane recommends that searches for intervention reviews maximize sensitivity while striving for reasonable precision. Increasing sensitivity commonly decreases precision because broader retrieval brings additional irrelevant records along with the relevant ones.
This is why a comprehensive search can generate thousands of records that are later excluded. The screening burden is not necessarily evidence that the search was poorly designed. Sometimes it is the cost of reducing the risk of missing eligible studies.
Systematic and Comprehensive Often Work Together
A systematic review usually wants both.
The search should be systematic so that evidence identification follows a planned and transparent process rather than an arbitrary sequence of discoveries. It should also be sufficiently comprehensive so that the evidence base is not unduly shaped by easily found studies.
Cochrane explicitly describes its approach using both terms: systematic and comprehensive.
This pairing is informative. If the concepts meant exactly the same thing, there would be little reason to specify both.
A search can follow a systematic procedure yet be constrained in comprehensiveness because of limited database access, language capability, time, resources, or deliberate scope restrictions. Those constraints should be acknowledged rather than hidden beneath the word “systematic.”
Systematic Searching Usually Requires Reproducibility, but the Concepts Remain Distinct
Reproducibility is normally an important property of systematic searching because readers need to inspect how evidence entered the review.
Gusenbauer and Haddaway explicitly include transparency and reproducibility among the defining goals of systematic searching. Cochrane likewise requires search documentation sufficient for reproducibility to the extent possible.
Still, the concepts are analytically distinct.
You can reproduce a non-systematic search if it was documented carefully. For example, a researcher might record an exploratory Google Scholar query, the date, and how the first 100 results were examined. The procedure is documented, although the search was not necessarily systematic and Google Scholar's changing ranking limits exact reproducibility.
Conversely, researchers may describe their approach as systematic but report too little information for another researcher to reconstruct what was actually done. The label does not create reproducibility after the fact.
A Search Can Be Extensive Without Being Systematic
Imagine a researcher who spends several months searching many databases, browsing journals, following citations, checking reference lists, consulting colleagues, and searching the web.
The researcher may identify an impressive body of literature.
But if the process developed opportunistically, the boundaries remained unclear, and the procedures were not documented, the search may be extensive without being systematic or reproducible.
This does not necessarily make the work useless. Exploratory and narrative forms of searching can legitimately evolve through discovery.
The problem appears when the researcher later claims that the search systematically identified the relevant evidence because a great deal of searching happened.
Effort and methodological properties are not the same thing.
The Same Search Can Be Systematic, Reproducible, and Comprehensive
These distinctions are not intended to force you to choose one label.
A well-designed systematic-review search may reasonably aim to possess all three qualities.
Systematic The search follows predefined eligibility criteria and a planned evidence-identification process across appropriate sources.
Reproducible The databases, platforms, dates, complete search strategies, limits, filters, and supplementary methods are documented and reported sufficiently for the procedures to be reconstructed.
Comprehensive The search uses appropriate databases, sufficiently sensitive terminology, and complementary discovery methods to identify as much eligible evidence as reasonably possible.
The three qualities reinforce one another. Systematic planning makes comprehensive coverage more deliberate. Reproducible documentation makes the systematic process inspectable. Comprehensive retrieval reduces the risk that the resulting synthesis represents only an easily discoverable subset of the evidence.
But none should be assumed merely because another is present.
The Appropriate Standard Still Depends on the Kind of Search You Are Conducting
Not every research search needs to maximize all three qualities.
If you are conducting an exploratory search to understand an unfamiliar topic, comprehensive coverage may not yet be the objective. You may deliberately follow promising concepts, learn terminology, and change direction as your understanding develops.
If you are trying to relocate one known paper, reproducible documentation of a multi-database strategy would be absurdly disproportionate.
If you are conducting a systematic review, however, the standards change substantially because the search determines which evidence becomes available for synthesis.
The useful principle remains proportionality. The methodological properties of the search should match the research task and the strength of the claims you intend to make.
That is ultimately why search quality has to be judged in relation to the kind of study being conducted.