How to Onboard a New Collaborator to a Literature Review

Onboarding a new collaborator to a literature review means transferring not just a folder of PDFs but a working method: the eligibility criteria, the search logic, the screening decisions already made, and the evidence trail that lets the newcomer reproduce and extend the work. The practical goal is a handoff package that a competent colleague can pick up, run one search, screen one batch, and reach the same conclusions you would.

This guide gives you a sequence with checkpoints, a worked example, a reusable checklist, and clear rules for when to stop or escalate. It assumes you are the person who owns the review and are bringing someone in mid-stream — for a manuscript revision, a coauthor addition, a peer-review response, a lab handoff, or the start of a new review cycle.

Most failed handoffs happen because the owner assumes the collaborator can see the method that lives in their head. Before you draft anything, audit what actually exists on disk and in your notes. Ask yourself four questions.

1. Is there a written protocol? Cochrane's guidance on searching and selecting studies begins with planning searches from explicit eligibility criteria. If your criteria exist only as intuition, the first onboarding task is to write them down, not to explain them verbally.
2. Can your searches be re-run? A search is reproducible only if the exact terms, fields, filters, and sources are recorded. PubMed's search details show how terms were translated and combined, which is exactly the record a newcomer needs. If you cannot reconstruct your last search, neither can they.
3. Are screening decisions documented? Cochrane recommends screening in stages and linking multiple reports of one study. If you have been screening informally, the collaborator cannot tell which records were excluded or why.
4. Is there a single place where papers live? Scattered downloads, email attachments, and browser bookmarks are not a library. Consolidate before you onboard.

Write a one-paragraph "state of the review" note answering these four questions. That note becomes the cover page of your handoff package and tells you which gaps to close first.

A handoff without a defined scope produces either idle collaborators or duplicated work. Decide, in writing, which of these roles the newcomer takes:

• Search runner: re-runs and extends searches, records search strings and dates.
• Screener: applies eligibility criteria to titles, abstracts, and full texts, logging decisions.
• Evidence mapper: links multiple reports of the same study and tracks which papers support which claims.
• Citation auditor: verifies that references in the draft correspond to real, correctly described sources.
• Writer: drafts synthesis sections from the screened evidence set.

For a first onboarding, assign one primary role and one shadowing role. A newcomer who both runs searches and writes synthesis in week one will produce work you cannot audit. Match the role to the artifact you can hand over most cleanly.

A literature review is reproducible when its decisions can be inspected. Assemble these artifacts into one shared location and name them consistently.

State inclusion and exclusion criteria in operational terms: population, intervention or topic, study design, date range, language, and publication type. Cochrane's approach ties search planning directly to these criteria, so write the criteria first and derive search concepts from them.

For every source, record the database or tool, the exact query string, field tags or filters used, the date run, and the number of records returned. PubMed supports field tags, phrase searching, filters, and saved searches, and its search details reveal how terms were translated and combined — capture that translation in your log. If you also use Google Scholar, note that it supports searching by author and title, date limits, and citation export, so record which of these you applied.

A simple table with columns for record identifier, stage (title/abstract or full text), decision, reason for exclusion, and reviewer initials. Cochrane recommends staged screening and deduplication; your log should show both.

A table linking each included study to the claims it supports, and noting when multiple reports describe one study. This is the artifact that prevents a newcomer from double-counting the same cohort published twice.

Export citations in a standard format and keep them in one file. If you maintain a BibTeX file, it can be screened against real academic databases with TrueCite, which flags references that do not correspond to real records. Run this check before the collaborator starts writing so they are not building on a corrupted bibliography.

Decide where papers, notes, and logs live, and give the newcomer access before your first meeting. If you want a single place to save and upload papers and organize them, My Library lets you build that collection. Pair it with a shared document for the protocol, search log, and screening log so the method trail is not trapped inside one person's account.

For a review that will run across weeks, create a dedicated space with Research Projects so the collaborator can see the project rather than a pile of links. The rule is simple: every artifact in the handoff package has exactly one canonical location, and the newcomer is told where it is.

Do not describe the method abstractly. Run one search together, live, and narrate every decision.

1. Restate the question and criteria. Point to the protocol line that justifies the next step.
2. Build the query from concepts. Show how free-text terms combine with subject headings, as Cochrane recommends, and how field tags narrow the search in PubMed.
3. Run it and read the result count. Open the search details and show how the system translated the terms.
4. Export and deduplicate. Save the records and remove duplicates before screening.
5. Screen a small batch together. Apply the criteria aloud, record decisions in the screening log, and show where a borderline record gets escalated to you.

If you use WisPaper's scholarly search, run this walkthrough in WisPaper search so the collaborator learns the interface you will actually use. The point of the live session is not speed; it is that the newcomer sees the reasoning behind each recorded decision.

Checkpoint: the collaborator independently re-runs the same search and produces the same record count and the same first ten screening decisions. If they do not, the log is incomplete — fix the log, not the person.

Screening is where onboarding most often drifts, because borderline judgments are where tacit knowledge lives. Give the collaborator three rules:

• Include when the record clearly meets every criterion. Log the decision and move on.
• Exclude when a stated criterion is clearly violated. Log the reason using the protocol's wording, not a free-text paraphrase.
• Escalate when the record is ambiguous, when two reports may describe one study, or when a criterion does not cover the case. Escalation means flagging the record for you with a one-line question, not guessing.

Set a batch size for the first week — for example, the collaborator screens a defined batch and you review every decision before they continue. Once agreement is consistent across two batches, move to spot-checking. Cochrane's emphasis on staged screening and on linking multiple reports of one study gives you the vocabulary to explain why the escalation rule exists.

Once the collection is assembled, a collaborator will have questions: which papers address a sub-question, what a study actually reported, where a claim came from. Answer these against the shared collection rather than from recollection. Library QA supports asking questions based on papers uploaded to My Library or added from search results, which keeps the answers anchored to the evidence set the whole team shares.

For longer stretches of work — drafting a synthesis, exploring an adjacent question, or planning a reproduction — Scholar Agent covers literature review, inspiration, and reproduction use cases. Assign it a defined task with a defined output so the collaborator's work remains checkable against the protocol.

Suppose you are revising a manuscript and a new coauthor joins to strengthen the related-work section. Here is the sequence in practice.

Step 1 — State of the review. You write: the review question, the eligibility criteria, the two databases searched, the date of the last search, and the fact that screening was done by one person without a written log. The gap is obvious: no screening log.

Step 2 — Scope. The coauthor owns the citation audit and the related-work synthesis; you retain search execution and final inclusion decisions.

Step 3 — Package. You write the criteria into the protocol, reconstruct the search strings from PubMed's search details, and build a screening log for the records already included. You export the reference file and screen it with TrueCite to catch references that do not resolve to real records.

Step 4 — Workspace. Papers move into My Library; the protocol and logs live in a shared project created in Research Projects.

Step 5 — Live search. You re-run the search together, export, deduplicate, and screen ten records aloud.

Step 6 — Screening handover. The coauthor screens the next batch and escalates ambiguous records. You review every decision in batch one.

Step 7 — Synthesis. The coauthor drafts the related-work section from the evidence map, citing only records in the included set.

Step 8 — Pre-submission check. Cochrane recommends re-running searches before publication. You re-run both searches, log the new date and counts, screen any new records, and update the evidence map before the revision goes back to the journal.

The handoff is complete when the collaborator can do three things without you: re-run a logged search and get the same count, screen a batch and log decisions that match your judgments on the spot-check, and trace any claim in the draft back to an included study through the evidence map.

Stop and escalate to a fuller reset when any of these occur:

• The search cannot be reconstructed. Rebuild the search log before the collaborator runs anything new.
• Screening agreement stays low after two reviewed batches. Return to the protocol and rewrite the criteria that are causing disagreement.
• The reference file contains entries that do not resolve to real records. Screen the file and correct it before synthesis continues.
• Two reports of one study have been counted separately. Rebuild the evidence map for the affected studies.
• The collaborator is asked to make a final inclusion decision alone. Final calls stay with the review owner until the escalation rules have been tested across several batches.