How Do You Document Search Changes Without Pretending They Were Planned From the Beginning?

What Should You Report When Your Final Search Is Not the Search You Started With?

You begin with one set of keywords. After screening several hundred records, you discover an important synonym. A database's subject headings work differently from what you expected. Citation searching reveals terminology from another discipline. One concept produces mostly irrelevant results, so you refine it.

By the end, your search is considerably better than version one.

The temptation is to report only the polished final strategy, as though you had designed it perfectly before seeing the first record. That creates a cleaner methods section, but it can erase information that matters for understanding how the evidence base was constructed. Transparent documentation should preserve the difference between what was planned and what was learned during the search.

Keep a Search History That Separates Plans From Later Decisions

What a Transparent Search Record Should Preserve

Search documentation and search reporting are related but different

Cochrane makes a useful distinction between documenting a search and reporting it. Documentation is the internal record you maintain while conducting the review. Reporting is the information eventually presented to readers.

Good reporting depends on good documentation because memories of exactly which terms, filters, dates, platforms, and iterations were used become unreliable surprisingly quickly. Search interfaces also change, and databases may use different syntax even when the underlying concepts are the same.

For Cochrane intervention reviews, documenting the search process in enough detail to permit correct reporting is a mandatory methodological expectation. The guidance specifies that records should capture the sources searched, when they were searched, who conducted the search, and which terms were used, with searches reproducible to the extent possible.

Preserve the exact searches you actually ran

Do not reconstruct a search months later from memory if the original database output or search history can be retained.

For each important database or information source, useful records may include the database name, platform, date searched, exact search strings or line-by-line strategy, controlled vocabulary, fields searched, filters, limits, and number of records retrieved. What is appropriate varies across databases and review methods.

Cochrane recommends copying and pasting or exporting exact database strategies rather than retyping them, partly to reduce transcription errors. PRISMA 2020 asks systematic-review authors to present full search strategies for all databases, registers, and websites searched, including filters and limits.

That final reproducible strategy is essential. It is not always the whole history.

Keep versions when the strategy changes materially

If an early exploratory search is discarded after five minutes because of a spelling error, maintaining an elaborate methodological diary would add little value. Documentation should be proportionate.

More consequential changes deserve a record. Examples include adding or removing a major concept, substantially changing terminology, introducing a date or language restriction, changing databases or other information sources, adding a search method such as citation searching, or revising a strategy because known relevant studies were being missed.

State why the search changed, not merely that it changed

“Search terms were refined during the review” provides little information. What did you learn that required refinement?

Perhaps screening showed that an established synonym had been omitted. Perhaps a subject heading had changed. Perhaps a term was so ambiguous that it retrieved large numbers of irrelevant records. Perhaps citation searching identified terminology used in an adjacent discipline. Perhaps the search failed to retrieve known eligible studies.

These explanations make the methodological logic visible.

They also help distinguish legitimate iteration from changing search terms in ways that could introduce confirmation bias.

Do not rewrite chronology

Suppose your original strategy omitted an important synonym. After screening, you discovered the term and reran the search with it included.

A final methods section that presents only the improved strategy without any indication that the search developed may be adequate for some limited reporting purposes, depending on the review type and what exactly is being reported. But where the change represents a consequential departure from a protocol or materially affects how the evidence was identified, describing the revision gives readers information needed to evaluate the process.

The aim is not confession. Iterative refinement can be methodologically appropriate. The issue is whether the written record accurately represents what happened.

Planned methods and actual methods should be distinguishable

Protocols and registrations provide a useful reference point because they record intended methods before the review is completed. PRISMA 2020 includes an item asking authors to indicate where a protocol can be accessed and to describe and explain amendments to information provided at registration or in the protocol.

If the final search differs from the planned search, identify the relevant difference rather than silently replacing the planned method with the final one.

A rigorous report can contain both. They are not competing versions of the truth.

Document sources beyond bibliographic databases too

Literature identification may involve more than database searching. Reference-list searching, forward citation searching, trial registers, websites, organizations, expert contacts, and other sources can contribute relevant records.

Cochrane specifically notes the importance of planning how searches of sources other than databases will be recorded because activities such as reference-list searching and forward citation searching may occur later in the review process. Decisions about search iterations can also be preserved in appendices and used in future updates.

For an iterative search, this matters because later search methods can otherwise become invisible. A reader may see a final set of included studies without knowing that some were discovered through a substantially different route from the original database search.

Record restrictions and filters carefully

Date, language, document-type, study-design, and other restrictions can materially change what a search retrieves. When such limits are used, their rationale may matter as much as the syntax.

Cochrane advises review authors to justify restrictions on publication date and format and cautions against language restrictions because of the risk of excluding relevant studies. PRISMA 2020 asks authors to report filters and limits as part of the full search strategy.

If a restriction was added after searching began, preserve when and why it was introduced rather than allowing it to appear retrospectively inevitable.

Do not document so little that the search cannot be reproduced

Statements such as “Google Scholar and several databases were searched using relevant keywords” are not reproducible descriptions.

For systematic reviews, readers should be able to see which sources were searched, when they were searched, and the actual strategies used. Search reproducibility is imperfect because database content and interfaces change, but that does not remove the obligation to document what was done at the time.

Do not document so much that the important decisions disappear

The opposite problem is possible too. A search diary containing every typo, test query, and minor syntax adjustment can become enormous while obscuring the decisions that actually shaped retrieval.

A useful internal record can be detailed, while the published explanation focuses on consequential changes. Supplementary materials can carry full strategies and technical detail when the main article cannot reasonably accommodate them.

Good documentation makes later updates much easier

A clear search history is useful long after publication. When a review is updated, previous strategies may need to be rerun or modified because database indexing, controlled vocabulary, interfaces, or understanding of the original search has changed.

Cochrane's guidance on review updates recognizes that experience from the original search may justify modifying sources or search terms. Without an accurate record of the original search and its development, updating becomes partly an exercise in methodological archaeology, which is rarely anyone's favorite research method.

How to Record an Iterative Search Without Rewriting Its History

Nothing embarrassing occurred in this example. The search improved. The methodological record simply preserves how it improved.

Common Mistakes When Reporting Search Changes

Keep a Small Search Audit Trail From the Beginning

You do not need sophisticated software to create a useful search history. A structured spreadsheet, research log, reference-management note, or review platform can work if it preserves the necessary information consistently.

Be especially transparent when a search modification changes which kinds of evidence become visible. If a revision was made after you had already seen the direction of findings, document the rationale carefully and consider whether the change could have been influenced by those findings.

The same principle applies beyond search terms. If your inclusion criteria changed after results became visible, preserve that methodological history as well.

Ultimately, a good audit trail helps you answer a deceptively simple question: could another researcher understand not only what search you ended with, but how the consequential decisions that produced it were made?

What to Record When Your Literature Search Changes

Frequently Asked Questions About Search Documentation

Report the Search You Actually Conducted, Including How It Evolved

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.