Tabular Review Overview
Tabular Review transforms unstructured legal documents into filterable, analyzable datasets. Documents become rows; extracted data points become columns. Each extracted answer traces back to its source through citations. This is designed for due diligence, portfolio reviews, compliance audits, discovery, and any workflow where you need structured comparison across many documents.
This video shows how to create a Tabular Review from Assistant.
How Tabular Review Works
Add documents: Upload files or select folders from your workspace.
Define columns: Specify what to extract from each document. Choose response types like text, yes/no, number, date, or select options.
Run extraction: August analyzes each document and populates cells with extracted values.
Verify with citations: Each cell links to the source passage. Click to confirm the extraction matches the document.
Export and use: Download as Excel or CSV for downstream analysis, or continue refining in August.
This video shows how to build a tabular review from scratch.
Two Ways to Build a Review
Choose the method that fits your workflow:
Method | How It Works | Best For |
|---|---|---|
Via Assistant | Describe the extraction task in natural language; August proposes columns automatically | Fast setup, exploratory extraction |
Manual Setup | Define columns, response types, prompts, and document scope through dedicated interface | Precise control, repeatable schemas |
Path A: Via Assistant
Attach docs
Upload or connect the documents you want to analyze.
Describe in English
Tell August what to extract in plain language.
August proposes columns
AI suggests column structure based on your description. Click Create tabular review on the approval card to build it, or Answer in chat to get the answer as a chat reply instead.
Populated table
Review and refine the populated table.
Path B: Manual Setup
New Review
Create a new Tabular Review from scratch.
Define columns + types
Add each column, specify response type, and write extraction prompts.
Select scope · run
Choose documents and run extraction.
Populated table
Review results and iterate as needed.
Both paths converge on the same interactive table output. You can also start from the Folders page: select files or folders and choose Send to Tabular Review to add them to an existing review or click Create New Review.
This video explains how citations work in Tabular Review.
Column Types and Response Formats
When building a Tabular Review, each column extracts a specific type of information. Choose the response type that matches the data you're extracting:
Type | Description | Example Use |
|---|---|---|
Text | Free-form text output | Party names, clause summaries, descriptions |
Yes/No | A yes or no answer | Does the contract contain an arbitration clause? |
Number | Numeric values | Contract value, liability caps, term length |
Date | Date values in ISO format | Effective date, expiration date, renewal dates |
Single Select | One option from a predefined list | Governing law jurisdiction |
Multi Select | Multiple options from a predefined list | IP types covered |
Verbatim Quote | Exact language from source document | Specific clause text, defined terms |
Auto | August chooses the most suitable format | When you want the system to infer the best type |
For detailed response type guidance, see Column Types and Response Formats.
Work With Columns and Cells
Open a column header's menu to sort, filter, or change the column:
Sort Ascending or Sort Descending orders the rows by that column.
Filter by value narrows the rows; it's available on Yes/No, Single Select, and Multi Select columns.
Duplicate copies the column with its answers, and Delete removes it.
Column Comments opens the discussion on that column.
Drag a column header to reorder columns.
To re-extract a column, edit its prompt or settings and click Update. To correct a single answer yourself, open the cell and use Edit Cell.
Per-File vs. Per-Folder Analysis
Per-file analysis: Evaluates each document independently, typically one row per file. Use when each document should be treated as a separate unit.
Folder-level analysis: Treats related documents as one unit. Use when answers depend on cross-document context (agreements with exhibits, amendments, or related correspondence).
Choose analysis mode based on whether context spans multiple files. This decision materially affects extraction behavior and interpretation.
When you add a folder, choose Evaluate as folder (all files together as one row) or Evaluate per file (each file as its own row). Check Apply to all future folders in this review to skip the question next time. From the Folders page, the same choice appears as As folder or Per file.
Templates and Reuse
Create templates from Tabular Review and reuse them across document sets:
A template stores recurring questions/columns so teams avoid rebuilding the same structure.
Save any review as a template for future use.
Template reuse improves consistency across matters.
Use clear template naming to support collaboration. Consistent column typing when creating templates ensures columns and options are preserved.
Exporting Template Prompts
Export a template's column prompts as a spreadsheet when you want to review or share its configuration outside August:
In a Tabular Review, click Templates.
Click the download icon (Download prompts) on the template card.
The Excel file is named <template-name>-prompts.xlsx and has columns for Name, Type, Prompt, Options, Web Search, and Ref Doc Count. A template with no columns can't be exported.
Exporting and Downstream Use
Click Download in the review toolbar to export the results:
Download as Excel (.xlsx)
Download as CSV for import into other systems
Check Include Citations in the same menu to include source citations in the export.
Exported outputs are useful for partner updates, diligence summaries, and downstream drafting. Before downstream use, validate key fields against cited source text. Use filters to narrow the view before export for issue-specific reporting.
To keep working on the results in conversation, attach the review in Assistant with Add tabular reviews. August can answer questions across the whole table and add or change columns.
Collaboration and Notifications
Work together in real time on Tabular Reviews:
Share reviews: Share a tabular review with colleagues for collaborative review. Only people who can edit the review can share it.
Project access: owners and editors of a project the review belongs to can edit it, even when it wasn't shared with them directly. Project viewers can view it.
Move or copy: use Move to Project or Copy to Project in the reviews list to put a review in another project.
Comments: Collaborators can comment on individual cells to ask questions or flag issues.
Review status: Mark each cell as verified, needing review, or incorrect to keep a clear review trail.
When someone shares a review with you, you get a notification in August. See Collaboration and Notifications.
Delete Reviews
Remove Tabular Reviews you no longer need from the reviews list. Deleted reviews disappear immediately and cannot be reopened. See Delete Tabular Reviews for the deletion workflow and how it affects shared reviews.
When to Use Tabular Review
Due diligence: Extract key terms from NDAs, MSAs, and purchase agreements across a deal room.
Portfolio reviews: Compare lease terms, liability caps, or renewal dates across a document set.
Compliance audits: Check for required clauses, signature blocks, or regulatory language across contracts.
Discovery: Organize extracted facts from productions for case analysis.
How It Differs from Assistant
Use Tabular Review when you need structured comparison across many documents. Use Assistant when you need conversational analysis of one or a few documents.
Tabular Review is built for repeatable, large-scale extraction where consistency matters. Each column follows the same definition across all rows, enabling reliable comparison.
Size Limits and Large Document Sets
Each operation on a Tabular Review can process up to 20,000 new cells. We count new cells as documents × columns before extraction starts, and decline requests that exceed the limit with a This review is too big message that shows the documents and columns counted.
The limit applies separately to each operation: creating a review, adding documents, adding a column, or re-running a column. For example, adding 500 documents to a review with 50 columns processes 25,000 cells and is declined, while adding 300 documents (15,000 cells) succeeds.
If your extraction task hits the cell limit, add documents in smaller batches, create multiple smaller reviews, or remove non-essential columns.
What counts as a cell
A cell is one extracted value for one document. A review with 200 documents and 25 columns produces 5,000 cells (200 × 25). Adding more documents or more columns increases the cell count proportionally.
A folder evaluated as one row still counts every file inside it, including files in its subfolders. A file that appears twice in the same request counts once, and cells already in the review don't count toward a new operation.
Working within the limit
If you need to analyze a large corpus:
Add documents in batches: Each batch is checked on its own.
Split into multiple reviews: Create separate reviews for subsets of documents (by folder, date range, or document type).
Reduce column count: Extract only the most critical fields in the initial pass.
For very large extractions that cannot be split, contact your account manager to discuss alternative approaches.
What to Read Next
Column Types and Response Formats for response type reference
Delete Tabular Reviews to remove reviews you no longer need
Assistant Overview for creating reviews from chat