Definitions drift
A field such as context or effectiveness can mean something different in every paper when its extraction rule is implicit.
Extract the fields your research question actually needs.
Define each field by name, extraction instruction, and evidence requirement. EvidenceGrid applies the same schema across selected PDFs and keeps every substantive result attached to a quote and page.
Free includes 10 papers per month, the complete evidence workflow, and portable exports.Workflow
Build a repeatable extraction schema in three steps.
Use one clear definition for every paper, then review exceptions at field level.
- 01
Frame the research question
State the comparison or explanation the matrix must support so every field has a clear purpose.
- 02
Define fields and evidence
Add up to twelve custom fields, with a focused extraction instruction and an explicit evidence requirement for each.
- 03
Review and reuse the structure
Check exact quotes and pages, repair a single field when needed, then save the schema as a template or copy the workspace structure.
A generic spreadsheet rarely survives a real review.
Discipline, method, and research question change what counts as useful evidence. A fixed list of columns often creates blank cells, inconsistent interpretation, and another round of manual cleanup.
Evidence gets separated
A concise matrix value is difficult to verify after the supporting text and page have been copied somewhere else.
The schema gets rebuilt
Related projects often begin by recreating the same research question, field descriptions, and evidence rules.
Inside EvidenceGrid
A field is more useful when the evidence rule is explicit.
Custom extraction columns separate the value you want from the instruction used to find it and the source evidence required to accept it.
Custom extractionWhat explains successful human and AI collaboration?
01Theoretical framework
Identify the explicit theory or conceptual framework.
Require a passage that names or defines the framework.
02Key theme
Extract the central theme relevant to the research question.
Require a passage that develops the theme.
03Study limitation
Extract only limitations explicitly stated by the authors.
Return Not explicitly reported when no source passage exists.
Keep the schema stable and the exceptions visible.
EvidenceGrid gives each field a consistent rule while preserving the review controls needed for ambiguous papers.
Up to twelve custom fields
Name the concept, describe how to extract it, and define what source passage counts as acceptable evidence.
Built-in and personal templates
Start with narrative, systematic, qualitative, or intervention review fields, or save a custom structure under your own name.
Copy workspace structure
Reuse the research question, field instructions, and evidence requirements without copying papers or prior results.
Single-field repair
Reprocess one uncertain cell while leaving accepted evidence and the rest of the matrix unchanged.
Plans
Test the complete extraction workflow before the review grows.
Free includes custom fields, saved templates, copied workspace structures, evidence review, and 10 papers per month. Researcher raises capacity to 120 papers per month and adds one-click Zotero and Notion delivery.
Compare Free and ResearcherQuestions researchers ask.
What information can I extract from a research paper?
You can define bibliographic, methodological, theoretical, contextual, outcome, limitation, or discipline-specific fields. A field should serve the research question and include a clear evidence rule.
How many custom extraction fields can I add?
A workspace supports up to twelve custom fields. Keeping the schema focused makes comparison and source review easier.
Can I reuse the same extraction form in another project?
Yes. Save the custom fields as a named extraction template, or copy a workspace structure to preserve the research question and field definitions without copying papers or results.
What happens when a paper does not report a field?
The extraction instruction can require an explicit missing-data response such as Not explicitly reported. Review states and source matching keep that absence visible.