03 · What You Need to Know
The Right Starting Point Depends on How Much Direction You Already Have
Researchers sometimes ask whether a literature search should be broad or narrow as though there must be one generally superior strategy.
There is not.
The useful question is: How much uncertainty do you currently have about what you are looking for?
If you do not yet know the relevant terminology, conceptual boundaries, major theories, or even exactly what problem deserves investigation, a highly focused search may encode assumptions you have not yet had an opportunity to question. If the research question and eligibility criteria are already well specified, repeatedly wandering through broad searches may instead create unnecessary noise.
This is why the scope of searching should follow what you need the literature search to accomplish.
Broad Searching Is Useful When You Are Mapping an Unfamiliar Territory
A broad exploratory search is particularly useful near the beginning of a project when you are trying to understand how a field talks about your topic.
You may begin with terminology drawn from everyday language, professional practice, media discussion, or another discipline. Researchers may use different terms. Several distinct constructs may exist beneath what initially appeared to be one concept. Alternatively, several labels may describe essentially overlapping phenomena.
Broad searching gives you opportunities to discover those differences before you commit to a narrow vocabulary.
For example, someone interested in students “depending too much on AI” might initially search that phrase. Relevant scholarship, however, could discuss concepts such as overreliance, automation bias, cognitive offloading, help-seeking, metacognition, or technology dependence, depending on the phenomenon actually being studied. Those terms are not necessarily interchangeable, but encountering them helps the researcher understand how the initial concern might connect with established scholarly conversations.
The purpose of the broad search is therefore not simply to retrieve more records. It is to learn how to search more intelligently.
Broad Does Not Mean Searching Without Boundaries
An exploratory search still needs a reason for existing.
Searching “education” because you are interested in an educational problem is not usefully broad. It is largely unbounded. Even an exploratory search should normally have a conceptual center: a phenomenon, population, setting, problem, intervention, technology, theoretical idea, or relationship that provides some direction.
Think of breadth as allowing uncertainty around the boundaries of the problem rather than abandoning boundaries altogether.
Broad exploratory search
Uses deliberately flexible boundaries to discover terminology, concepts, adjacent literatures, major debates, and possible directions relevant to an emerging problem.
Unfocused searching
Retrieves material without a clear information need, stopping logic, or basis for deciding which discoveries are actually relevant to the research problem.
The difference is intent. An exploratory researcher may not yet know exactly where the inquiry will end, but should still know what they are trying to learn from the search.
A Focused Search Becomes More Useful as the Question Becomes Clearer
Once you understand the topic better, you can begin tightening the search around the concepts that matter to the emerging research problem.
This usually improves efficiency. Literature that was useful during orientation may no longer be directly relevant to the study you are developing. Your attention shifts from learning the landscape to examining the scholarship that can actually inform, challenge, or justify the proposed research.
Suppose your broad interest in generative AI and higher education eventually becomes a question about how university students evaluate the accuracy of AI-generated formative feedback. Literature on institutional AI policy may still be interesting, but it is no longer central simply because it contains the words “generative AI” and “higher education.”
A focused question helps you decide what deserves sustained attention and what can reasonably remain outside the project's scope.
Focused Questions Can Improve Search Design Without Requiring Every Question Element to Appear in the Query
A well-developed research question can identify the concepts that a search strategy needs to represent. Frameworks such as PICO, commonly used for intervention questions in health research, can help translate a question into searchable concepts.
There is an important qualification, however. A detailed research question and a database search query are not the same thing.
Cochrane guidance notes that it may be unnecessary, and sometimes undesirable, to search every component of a clinical review question. Comparators and outcomes, for example, may be poorly represented in titles, abstracts, or indexing. Requiring those elements in a database query can cause relevant records to disappear from the results.
This illustrates a broader principle: a focused research question should focus your reasoning, but the corresponding search may need to remain relatively sensitive.
You might know exactly what evidence you want while deliberately using broad terminology within each important concept to increase the likelihood of retrieving it.
Broad and Focused Can Describe Different Parts of the Same Search
Another source of confusion is that “broad” and “focused” can refer to more than one thing.
Your research question may be focused while your search terms within a concept are broad. You may also search a broad range of sources for a narrowly defined type of evidence.
For systematic reviews of interventions, Cochrane recommends searches designed to identify as many eligible studies as possible. Within each search concept, a wide variety of synonymous or related terms may be combined to maximize sensitivity, while the overall search remains governed by explicit eligibility criteria.
| What can be broad or focused? |
Broad might mean... |
Focused might mean... |
| Research topic |
Generative AI in higher education |
Students' evaluation of AI-generated formative feedback |
| Research question |
How is generative AI affecting learning? |
A question specifying a particular phenomenon, population, context, or relationship |
| Search vocabulary |
Multiple synonyms, spelling variants, related terms, and controlled vocabulary for a concept |
A smaller set of highly specific terms |
| Evidence sources |
Multiple databases and supplementary discovery methods |
One specialized database or a deliberately bounded source |
| Eligibility |
Several populations, designs, contexts, or forms of evidence may qualify |
Clearly specified evidence must satisfy defined criteria |
Simply saying “use a broad search” is therefore incomplete advice. You need to know what is being broadened and why.
Early Broad Searching Can Help You Develop a Better Question
When your research problem is still emerging, broad searching can expose weaknesses in your initial framing.
You may discover that your proposed question has already been examined extensively. You may find that a phenomenon is normally conceptualized differently from the way you imagined. An apparently narrow problem may turn out to belong to a much larger theoretical debate. Conversely, a broad topic may contain one unresolved issue that is considerably more interesting than the original idea.
This is why you do not necessarily need a final question before beginning to search. Preliminary engagement with the literature can contribute to developing the question itself.
The key is to treat early searching as provisional. You are gathering information that will help you decide what deserves a more focused investigation.
Starting Too Narrowly Can Hide Relevant Terminology
A narrow search can create an illusion of scarcity.
Suppose you search a very specific phrase and retrieve six papers. It is tempting to conclude that little research exists. But perhaps the phenomenon has been studied extensively under another term. Perhaps your exact phrase became common only recently. Perhaps an adjacent discipline uses a different conceptual vocabulary.
This problem is particularly likely when you are new to a field or working with an emerging topic whose terminology has not stabilized.
Known relevant papers can help here. Examine their titles, abstracts, keywords, indexing terms, references, and terminology. Ask what language those researchers use that your original query did not.
Your initial search should teach you something about your next search.
Starting Too Broadly Has Its Own Costs
Broad searching is not automatically safer.
Every increase in breadth can increase the number of irrelevant records you must inspect. If your topic spans several mature literatures, an extremely broad search may generate thousands or tens of thousands of results without helping you distinguish what is central to your research problem.
This is partly an information-retrieval trade-off. Sensitivity, sometimes called recall, concerns retrieving relevant records. Precision concerns how much of what you retrieve is actually relevant. Increasing sensitivity commonly reduces precision.
For systematic reviews, accepting relatively low precision can be justified when missing eligible evidence would be more consequential than screening additional irrelevant records. Cochrane consequently advises maximizing sensitivity while striving for reasonable precision.
For an exploratory student project, however, retrieving 40,000 records merely because a search is “broad” may provide little intellectual benefit. Breadth should solve a problem, not become a badge of rigor.
Progressive Focusing Is Often More Useful Than Choosing One Extreme
For many research projects, the most practical approach is progressive focusing.
Begin with a research interest Identify the phenomenon, problem, population, or conceptual area that gives the search direction.
Explore broadly enough to learn Look for terminology, major concepts, influential scholarship, theoretical perspectives, related literatures, and obvious boundaries.
Develop a working question Use what you have learned to specify what you actually want to understand or investigate.
Search more deliberately Build searches around the important concepts in the emerging question, using terminology discovered during the exploratory stage.
Test and refine Examine whether the search retrieves known relevant papers and whether irrelevant results reveal ambiguities or unnecessary concepts.
This process can repeat. Search-strategy development is often iterative even after the question is well established. Cochrane guidance explicitly recognizes that search terms may be modified in response to what has already been retrieved.
Iteration, however, should become increasingly purposeful. Eventually, especially when the search is part of the research method, you need to move beyond exploratory querying toward a defensible search strategy built around explicit concepts and retrieval decisions.
Systematic Reviews Usually Start With a Focused Review Question but Still Use Broad Retrieval
Formal evidence synthesis requires an important distinction from early exploratory research.
By the time you execute the main search for a systematic review, the review's scope, question, and eligibility criteria should be defined sufficiently to determine what counts as potentially eligible evidence. Cochrane places determination of scope and questions before searching and selecting studies in its core methodological sequence.
Yet the search used to identify eligible studies may intentionally retrieve much more than the final included evidence set.
This is not contradictory. The decision problem is focused, while the retrieval strategy is sensitive.
Searchers may use numerous synonyms and controlled vocabulary terms, search several databases, and supplement database retrieval with other methods because relevant studies can be difficult to identify reliably. The goal is not exploratory wandering but broad capture within a defined methodological purpose.
Watch Out
Do not equate a narrowly written query with methodological rigor. For evidence synthesis, excessive search specificity can be dangerous if it prevents eligible studies from being retrieved. A focused review question and a highly restrictive database query are not the same thing.
Some Research Questions Require Iterative or Tailored Searching
Not every research problem can be reduced neatly to a conventional set of concepts and searched once.
Cochrane notes that frameworks such as PICO are not appropriate for every kind of question, including some questions involving qualitative evidence, diagnostic accuracy, prognosis, or methods. Searches for complex topics and some review approaches may require iterative strategies or tailored approaches.
This matters because advice to “focus the question and then search it” can imply a degree of stability that certain forms of inquiry do not possess.
For some questions, searching itself helps reveal the conceptual structure of the evidence. Multiple searches may address different aspects of a complex problem. Citation searching may expose bodies of literature that keyword searching does not retrieve effectively.
The appropriate response is not to remain permanently broad. It is to align the search architecture with the kind of question and evidence you are actually dealing with.
You Know It Is Time to Focus When Broad Searching Stops Changing Your Understanding
There is no universal number of searches after which exploration should end.
A practical signal is diminishing informational return.
If new broad searches continue to reveal important terminology, conceptual distinctions, major bodies of literature, or plausible alternative framings, further exploration may still be useful. If they mainly return variations of material you already understand, it may be time to concentrate on the question you have developed.
This is different from claiming that you have found every relevant publication. Rather, you have learned enough from exploration to make a defensible decision about what deserves focused investigation.
The threshold should become more demanding when the search itself forms part of the research method. In that situation, deciding whether searching is complete requires more than personal familiarity with the results and should be evaluated against the quality requirements of the particular study or review.