How Do You Know When Your Literature Search Is Too Narrow?

Does Getting Very Few Results Mean Your Search Is Too Narrow?

You run a literature search and retrieve 23 records.

That might feel reassuring. Twenty-three papers are manageable. You could probably examine every title and abstract before your next meeting, which is not something academics get to say often.

But is 23 evidence that your search is focused, or evidence that it is missing much of the relevant literature?

The result count alone cannot answer that question. A small result set may accurately reflect a specialized or emerging topic. It may also result from an overly restrictive phrase, too many concepts joined with AND, missing synonyms, an inappropriate database filter, incomplete controlled vocabulary, or search syntax that does not work as you intended.

A search becomes too narrow when its restrictions prevent it from retrieving relevant literature that it reasonably should be able to find.

A Search Is Too Narrow When Relevant Literature Is Being Excluded

A Small Result Set Can Mean Several Very Different Things

There is no minimum number of results a good search must retrieve

A search returning 30 records is not automatically too narrow, just as a search returning 3,000 is not automatically too broad.

The number of available studies depends on the maturity of the topic, disciplinary scope, database coverage, research question, study designs sought, publication period, and many other factors. A highly specialized question may genuinely have a small literature.

The difficulty is distinguishing a genuinely small evidence base from an artificially small result set created by the search strategy.

You should not infer the first merely from observing the second.

The clearest warning sign is a relevant paper you expected but did not retrieve

Suppose you already know of a paper that clearly addresses your topic. It is indexed in the database you are searching, yet your search does not retrieve it.

That is useful diagnostic evidence.

Cochrane recommends checking whether a developing search finds key publications or studies included in similar reviews as a basic test of whether the strategy is performing adequately. If a relevant record is present in the database but absent from your results, you can investigate exactly why it failed to satisfy the search.

Perhaps the authors used a synonym you omitted. Perhaps the relevant concept appears only in a subject heading. Perhaps your phrase is too exact. Perhaps one of your required concept blocks is not represented in the title or abstract.

A narrow search usually has low sensitivity, not merely few results

The concept most closely associated with excessive narrowness is sensitivity, also called recall.

This is why simply counting results is inadequate. A search returning 200 records might retrieve nearly all relevant literature available in the database, while another returning 2,000 could still miss important studies because of a poorly designed concept block.

For systematic reviews, Cochrane recommends maximizing sensitivity while striving for reasonable precision. The balance may differ for exploratory searches or conventional research projects, but the underlying principle remains useful: a smaller result set is not necessarily a better one.

Too many concepts joined with AND can suppress relevant literature

One of the most common ways to make a search too narrow is to require too many concepts simultaneously.

Suppose your research question concerns generative AI, academic writing, undergraduate students, writing self-efficacy, and online learning.

You might build:

generative AI AND academic writing AND undergraduate students AND writing self-efficacy AND online learning

A relevant paper can be excluded if it fails to satisfy only one of those blocks.

Perhaps the study involved undergraduates but the abstract says only university students. Perhaps it measured writing self-efficacy but refers to a named instrument rather than that phrase. Perhaps the course was online but delivery mode appears only in the full text.

None of those differences makes the study irrelevant. Yet the database may never return it.

This is why a complex question should first be broken into searchable concepts, followed by a separate decision about which concepts really need to become mandatory search blocks.

Outcomes and comparators can be particularly restrictive

A concept can be central to your research question while still being a poor mandatory retrieval condition.

Cochrane notes that it is usually unnecessary, and may sometimes be undesirable, to search every aspect of an intervention review question. Outcomes and comparators may not be well described in titles and abstracts and may also be inadequately represented by controlled vocabulary.

Suppose a review asks whether a teaching intervention improves critical thinking compared with conventional instruction. Requiring a critical thinking outcome block and a conventional instruction comparator block may seem faithful to the question. But relevant studies may describe the outcome through an assessment instrument and the comparator merely as a control group.

Removing a concept from the search does not remove it from the research question. It may remain part of the eligibility criteria and be assessed during screening.

The more fundamental decision is which concepts from the research question should actually become search terms.

Missing synonyms can create false scarcity

Searching only your preferred terminology assumes that relevant authors use the same vocabulary.

Consider the concept generative artificial intelligence. Depending on the literature and period, relevant records might use generative AI, GenAI, large language model, a specific technology name, or another appropriate expression.

If you search only one phrase, every relevant record using another formulation becomes harder or impossible to retrieve through that term.

Cochrane recommends using a wide range of appropriate free-text terms within selected concepts, considering synonyms, spelling variants, acronyms, truncation, and other suitable search features. Where databases use controlled vocabularies, relevant subject headings should also be considered.

The goal is not to accumulate synonyms indiscriminately. Each term should genuinely represent the intended concept in potentially relevant literature.

Controlled vocabulary and free-text terms can catch different records

Databases that assign controlled vocabulary offer another route to a concept beyond the words used by authors.

A record may use one expression in its title and abstract but be indexed under a standardized subject heading. Conversely, a newly published record may not yet have complete indexing, or an emerging concept may not map neatly to an established subject heading.

Relying only on controlled vocabulary can therefore miss useful text-word variants. Relying only on free text can miss records that indexing would have helped identify.

For comprehensive searching, Cochrane recommends combining appropriate controlled vocabulary with free-text terms and customizing the strategy to each database because controlled vocabularies and indexing practices differ.

An exact phrase can be narrower than the concept you mean

Quotation marks can be useful when several words need to occur together. They can also make a search unnecessarily restrictive.

Suppose you search only:

“generative artificial intelligence in academic writing”

A relevant paper using generative AI for student writing will not contain that exact phrase. Nor will a paper discussing large language models and academic writing.

The problem becomes much worse if you search the entire research question as one phrase. The database may then require wording that no relevant author has happened to use.

Phrase searching should represent meaningful linguistic units, not force the literature to reproduce your preferred sentence.

Overly restrictive field searching can hide relevant records

Searching only titles can produce a very focused result set because titles usually contain relatively few words. It can also miss studies where your concept appears in the abstract, keywords, or indexing rather than the title.

The same issue applies to other field restrictions. Restricting a term to a particular field can improve precision when the field is informative, but every field restriction also determines where the database is allowed to find the term.

If a known relevant article disappears after a field restriction, inspect where its relevant terminology actually appears.

PRESS guidance for peer review of electronic search strategies explicitly includes checking whether appropriate fields have been searched and whether keywords, index terms, concepts, and other elements are too broad or too narrow.

Truncation can be too narrow as well as too broad

Truncation is often used to retrieve multiple word endings from a common stem. But choosing the wrong stem can fail to capture variants you expected.

For example, a truncation may begin too late in the word to retrieve an alternative form, or the database may handle truncation differently from another platform you have used.

PRESS specifically treats overly narrow truncation as a search-quality issue alongside overly broad truncation.

Do not assume that a truncation symbol or wildcard behaves identically across databases. Check the current syntax for the platform being searched.

Database syntax errors can make a search silently narrower

A query can look reasonable to you while being interpreted differently by the database.

Potential problems include:

• using a truncation symbol from another search platform;
• incorrect Boolean nesting;
• placing quotation marks where the database handles them unexpectedly;
• using unsupported proximity syntax;
• searching an unintended field;
• combining search lines incorrectly;
• failing to map subject headings appropriately.

PRESS includes spelling, syntax, Boolean operators, proximity operators, subject headings, text words, limits, and filters among the major domains that should be checked when evaluating an electronic search strategy.

When a search behaves strangely, inspect the syntax before developing an elaborate theory about why the literature does not exist.

NOT can remove much more than you intended

The Boolean operator NOT can be useful in particular circumstances, but it deserves caution because a record is excluded if it satisfies the unwanted condition even when it also contains material you need.

Imagine trying to remove literature about secondary education with a NOT expression. A relevant article comparing university and secondary-school students might be eliminated because it contains the excluded term.

PRESS specifically asks reviewers to consider whether NOT is likely to cause unintended exclusions.

Filters and limits can create an artificially small literature

Date, language, document type, peer-review, subject, age, and other database limits can substantially reduce retrieval.

Some restrictions are methodologically justified. Others are merely convenient.

If your search retrieves suspiciously little, inspect every active filter. Ask whether each restriction follows from the research question, protocol, or eligibility criteria and whether the database applies it in the way you assume.

A five-year limit, for example, should be based on a defensible decision about the appropriate date range for the literature search, not used simply because older results feel inconvenient.

The same principle applies to decisions about language restrictions and whether to search only peer-reviewed literature.

One database can make the literature appear smaller than it is

Sometimes the search strategy is reasonable, but the information source does not cover enough of the relevant literature.

Databases differ in journal coverage, disciplinary emphasis, document types, indexing practices, and controlled vocabularies. A topic spanning education, psychology, medicine, computer science, and information science may not be represented adequately in a single disciplinary database.

That is a different problem from an overly narrow query, but the symptom can look similar: too few relevant studies.

If your search performs reasonably within one database yet obvious literature remains absent, consider whether database selection is contributing to the problem. The appropriate question then becomes how many databases you actually need to search, rather than endlessly broadening one query.

Check what disappears when you narrow and what appears when you broaden

A practical way to diagnose excessive narrowness is to compare versions of the strategy.

This is the mirror image of diagnosing a search that is too broad. In both cases, the objective is to understand what the strategy is doing rather than chase a preferred number of results.

Citation searching can expose missed literature

Once you have identified relevant studies, their reference lists and citing articles can provide another diagnostic route.

If citation searching repeatedly identifies relevant publications that your database strategy should have been capable of retrieving but did not, investigate why. Cochrane notes that finding many additional relevant records through citation searching may indicate that the original searches were not optimal and should be revisited.

This does not mean every citation-found article proves the database query was defective. Some records may not be indexed in the database you searched. Others may fall outside its coverage.

The useful question is whether the missed record was available to be found and, if so, why the strategy failed to find it.

Too few results should not be treated as evidence of a research gap

This is perhaps the most consequential mistake.

You search your exact topic and retrieve three studies. It is tempting to write that little research exists and declare a gap.

Before doing so, challenge the search.

Remove unnecessary concepts. Test alternative terminology. Examine controlled vocabulary. Relax exact phrases. Check filters and fields. Search other appropriate databases. Look at references and citations from relevant papers.

A search result is evidence about what a particular strategy retrieved from particular sources under particular conditions. It is not, by itself, evidence that nothing else has been studied.

Diagnosing Why a Search Finds Only a Handful of Papers

The important outcome is not that the search became larger. It is that the researcher could explain why relevant records had been excluded and modify the strategy accordingly.

If the revised strategy later produces excessive irrelevant material, that becomes a separate diagnostic problem. Search development often moves back and forth between sensitivity and precision rather than progressing neatly from “bad” to “perfect.”

Common Mistakes When a Search Retrieves Too Little

Broaden the Search by Removing the Restriction That Is Causing the Loss

When a search retrieves suspiciously little, resist the urge to rebuild everything immediately. Diagnose where relevant records are being lost.

A controlled change is more informative than indiscriminate broadening.

The objective is not simply to make the result count rise. A broader search is an improvement only when the additional retrieval increases your ability to find relevant literature at a cost in irrelevant retrieval that is acceptable for your purpose.

If you are developing the search from scratch, it can be useful to reconsider how broad the initial search should be. Starting with the central concepts and refining after inspecting the literature can make over-restriction easier to avoid.

Before Concluding That There Is Very Little Literature

Questions About Literature Searches That Retrieve Too Little

A Search Is Too Narrow When Its Restrictions Hide Relevant Literature

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.