You have a folder of PDFs, a half-finished reference list, and a supervisor comment that says "where did these come from?" You are not sure whether you searched enough, whether you screened consistently, or whether you could rebuild the same set of papers next month. The fix is not more searching. It is separating two records that most students collapse into one: the search log and the screening log.
A search log records what you searched, where, when, and how. A screening log records what you did with each record that came back: whether it was included, excluded, and why. Cochrane's guidance on searching for and selecting studies treats these as distinct planning and documentation tasks: searches are planned from eligibility criteria, run across multiple sources, and documented; records are then deduplicated and screened in stages. Keeping both logs makes your process reproducible and defensible. This article shows how to build each one, where each fails, and how to combine them into a workflow you can reuse for your next project.
What problem are you actually solving?
Most early-career researchers start with a search problem when they have a documentation problem. You may have run a database search, saved a few PDFs, and moved on. Later, when you need to write a methods section or respond to a reviewer, you cannot reconstruct which terms found which papers or why some records were dropped.
The two logs answer different questions:
- Search log: What did I do to find candidate records, and could someone repeat it?
- Screening log: What happened to each candidate record, and could someone audit my inclusion decisions?
Cochrane's handbook separates these concerns: search planning starts from eligibility criteria and uses multiple sources with free-text terms and subject headings; screening then proceeds in stages after deduplication. If you keep only one log, you lose either the ability to reproduce the search or the ability to justify the final set.
What belongs in a search log?
A search log is a record of your search process, not your reading list. It should let you or a collaborator rerun the same search and understand what changed between runs.
At minimum, record:
- Source: which database, index, or search interface you used.
- Date: when you ran the search.
- Query: the exact terms, field tags, filters, and limits you applied.
- Result count: how many records the search returned before you touched them.
- Notes: any translation of terms, phrase handling, or interface-specific behavior.
PubMed's search details, for example, show how terms were translated and combined, which is exactly the kind of information a search log should preserve. Google Scholar documents author and title searching, date sorting and limits, and citation export, all of which affect what your log needs to capture. Cochrane adds that searches should be planned from eligibility criteria, use multiple sources, combine free-text terms with subject headings, and be rerun before publication.
A practical format is a simple table with one row per search run. If you use a reference manager, keep the exported query or a saved-search link alongside the row. If you use a scholarly search interface such as WisPaper search, record the query and date there too, so the log reflects every source you actually used.
What belongs in a screening log?
A screening log is a record of decisions. It starts after deduplication and continues through title/abstract screening and full-text review.
At minimum, record:
- Record identifier: a stable ID or citation key.
- Stage: title/abstract or full text.
- Decision: include, exclude, or maybe.
- Reason: a short code tied to your eligibility criteria.
- Reviewer and date: who made the call and when.
- Notes: anything ambiguous that needs a second look.
Cochrane's process includes deduplicating records, screening in stages, and linking multiple reports of one study. Your screening log should reflect those steps: one row per record per stage, with a reason code that maps back to your protocol. If two reports describe the same study, note the link so you do not double-count.
A screening log is also where supervisor feedback becomes actionable. When your supervisor asks why a paper was excluded, the answer should be a reason code, not a memory.
How do the two logs compare on the same task?
Take a concrete task: you are writing a thesis chapter on a topic and need a defensible set of papers. Both logs serve the same project, but they answer different questions and fail in different ways.
| Criterion | Search log | Screening log |
|---|---|---|
| Purpose | Reproduce and document the search | Justify inclusion and exclusion decisions |
| Unit | One row per search run | One row per record per stage |
| Key fields | Source, date, query, filters, result count | Record ID, stage, decision, reason, reviewer |
| Supports | Methods section, rerunning searches, updating before publication | Audit trail, supervisor review, consistency checks |
| Fails when | Queries are not saved or sources are omitted | Decisions are made without reason codes or stages |
| Cochrane link | Planned from eligibility criteria, multiple sources, documented | Deduplicated, staged screening, linked reports |
Neither log replaces the other. A search log without a screening log tells you how you found records but not why your final set is what it is. A screening log without a search log tells you what you decided but not whether you found the right candidates in the first place.
Where does each log fail?
Search log failures usually come from informality. You run a quick query, save a few PDFs, and never record the exact terms. Weeks later, you cannot tell whether a missing paper was never found or simply never saved. Cochrane's emphasis on documenting the process and using multiple sources addresses this directly: the log is part of the method, not an afterthought.
Screening log failures usually come from skipping stages. You read titles and abstracts, then jump to full text without recording why records were dropped. When a supervisor or reviewer asks about a specific paper, you have no reason code. Cochrane's staged screening and deduplication steps exist to prevent exactly this.
A third failure is treating the logs as one artifact. If your search log doubles as your screening log, you lose the ability to separate search reproducibility from decision auditability. Keep them as separate files or separate sheets, even if they live in the same project folder.
What does a combined workflow look like?
A combined workflow moves from planning to searching to screening to updating, with each log capturing its own stage.
-
Plan from eligibility criteria. Write your inclusion and exclusion criteria first. Cochrane recommends planning searches from these criteria, so your search terms and screening reasons should map back to them.
-
Run and log searches. For each source, record the date, query, filters, and result count. Use multiple sources and combine free-text terms with subject headings where the source supports it. If you use a scholarly search interface, keep the query text in your log.
-
Deduplicate and export. Remove duplicate records before screening. Cochrane lists deduplication as a step before screening. Export the deduplicated set to your screening log.
-
Screen in stages. Screen titles and abstracts first, then full text. Record a decision and reason code at each stage. Link multiple reports of the same study so they are not counted twice.
-
Organize included papers. Save included papers in a library so you can return to them. My Library is where you can save or upload papers and organize them for later reading and questions.
-
Ask questions of your included set. Once papers are in your library, Library QA lets you ask questions based on papers uploaded to My Library or added from search results. This is useful for checking whether your included set actually covers the concepts in your criteria.
-
Rerun searches before submission. Cochrane recommends rerunning searches before publication. Your search log makes this possible: you can rerun the same queries and compare result counts, then screen only the new records.
-
Check citations before you submit. If you are preparing a reference list, TrueCite checks BibTeX against real academic databases to flag hallucinated references. This is a citation-screening step that runs after your screening log is complete.
For a larger project, Research Projects gives you a place to create and manage the project around these logs. If your next step is a longer literature review or a research task that spans weeks, Scholar Agent fits literature review, inspiration, or reproduction use cases.
Example: cleaning up a winter-break thesis chapter
Suppose you are finishing a thesis chapter over winter break. You have 60 PDFs, a messy reference list, and a supervisor meeting in January.
Starting state: You searched a few databases months ago, saved PDFs inconsistently, and cannot remember which terms found which papers. Your supervisor asked for a methods paragraph explaining how you selected studies.
Step 1: Rebuild the search log. For each source you remember using, record the date, query, and approximate result count. Where you cannot remember, note it and plan to rerun the search. Cochrane's guidance to document the process and use multiple sources supports this reconstruction.
Step 2: Build the screening log. Create one row per PDF with a decision and reason code tied to your chapter's eligibility criteria. Mark which records you have only seen as abstracts versus full text.
Step 3: Deduplicate and link. Remove duplicate records and link multiple reports of the same study. Cochrane lists both steps.
Step 4: Rerun searches. Rerun your logged queries and compare result counts. Screen only new records. This is the step that makes your January meeting productive: you can show what changed since the original search.
Step 5: Organize and check. Save included papers in My Library, ask coverage questions with Library QA, and check your BibTeX with TrueCite before you send the chapter.
Result: You have a methods paragraph backed by two logs, a defensible included set, and a rerun search that shows your chapter is current.
How do you know the result is ready?
Your logs are ready when someone else could repeat your process and reach the same included set. Check:
- Search log: Every source you used has a row with date, query, filters, and result count. You can rerun each query. You have noted any interface-specific behavior, such as how PubMed translated your terms.
- Screening log: Every record has a stage, decision, and reason code. Duplicates are removed. Multiple reports of the same study are linked.
- Combined: Your eligibility criteria appear in both logs: as search terms and as reason codes. You have rerun searches before submission and screened new records.
If a supervisor or reviewer asked "why is this paper in your review?" or "why was this paper excluded?", you should be able to answer from the logs without relying on memory.




