Legal Services Offshore research · legal-support-governance
Where does reviewer time accumulate in first-pass document review?
A time-observation design for first-pass document review that separates search, reading, coding, and exception handling without turning the result into a productivity ranking.

Research context
This research brief is designed for legal operations planning. Use the findings to ask better workflow questions. Test any proposed process against the firm’s jurisdiction, professional duties, systems, and client commitments. Treat every local result as a starting point for review rather than a universal benchmark.
Question and scope
This study asks where reviewer time accumulates during a first-pass document review. It treats the review as a set of observable stages: locating a document, reading it, applying a coding decision, and handling an exception. The scope is one review team, one document set, and a fixed observation window. The result describes that setting and should not be read as a benchmark for review speed.
The unit is one reviewer-document interaction that can be timed. A document opened and closed without a decision is still an event if it consumed review time. Documents that were never assigned are outside the scope. Record the inclusion rule before observing so the denominator is stable.
The study also records whether the reviewer worked from a search result, a folder, or a review platform. The tool affects search time and should be reported rather than pooled. If the team changed tools during the window, split the result by tool. A comparison across tools is only honest when the stage definitions stay the same.
Definitions and evidence standard
Define each stage before timing. Search time includes locating the document and confirming it is the right version. Reading time includes opening and reading the relevant pages. Coding time includes selecting the issue tags and writing the required note. Exception time includes stopping, flagging, and routing a document that the reviewer cannot decide.
Use a time record that the reviewer can complete without disrupting the work. A start and stop timestamp per stage is more reliable than a daily estimate. If the review tool logs activity, use the log and check a sample against the reviewer's note. Do not infer time from the number of documents alone. A slow document can be more valuable than ten routine ones.
Sampling and measurement
Select review sessions across different times of day and different document types. Freeze the sample before looking at the times. If the team uses more than one coding scheme, report the schemes separately. A session from a training day should be excluded or labeled, because it measures learning rather than routine work.
Measure the total time per stage, the median time per document, and the share of documents that required an exception. Add the number of documents that were reopened after coding. Report the count of documents and sessions behind each median. A median from five documents is a description, not a stable estimate.
Use a short observation form rather than a full time-tracking system. The form should take less than a minute to complete and should not require the reviewer to remember the start time. Test the form with one reviewer before the main observation. A burdensome form changes the work it is trying to measure.
Findings and interpretation
Present the stage distribution before suggesting a cause. A large search share may point to a finding aid or naming problem. A large reading share may be inherent in the document type. A large exception share may point to unclear coding rules. Read a sample of documents behind each stage before drawing a conclusion.
Compare the stages across document types and reviewer experience. A new reviewer may spend more time searching, while an experienced reviewer may spend more time on exceptions. Report the pattern with its uncertainty and avoid ranking individuals. The finding should suggest a specific support change, such as a better index or a clearer coding guide.
Compare the stage distribution with the number of documents in each session. A session with many short documents may spend proportionally more time searching and less time reading. Report the document count beside the median so the reader can see the mix. Do not compare sessions with very different mixes as if they were the same.
Limitations
Timing a review can change how a reviewer works. The observation may add a pause at each stage or encourage a reviewer to skip a careful check. Treat the result as a description of observed sessions, not a measure of unobserved performance. The study also cannot see the mental work that happens before a document is opened.
The document set may not represent the full review, and the window may include a training or a production surge. If the coding scheme changed during the window, the stages may not be comparable. Record these limits with the result. A time study is a planning tool, not a judgment about whether the review was adequate.
Record whether the reviewer took breaks or switched documents during a session. A session interrupted by unrelated work can inflate the measured time without reflecting the review itself. Note the interruption and exclude the session if it dominates the result. A transparent exclusion is better than a silent adjustment.
Practical next step
Choose one change that matches the largest avoidable stage, such as a document index, a coding guide, or an exception queue. Write the change, the owner, and the date it starts. Define the measure that will show whether it helped, using the same stage definitions. Test it on a comparable set of sessions.
Keep the before and after observations separate so the comparison is honest. Record the sessions that were excluded and why. Review the result with the reviewers and the lawyer who owns the review. If the change does not help, record that result too. The purpose is to learn where support reduces avoidable time, not to push reviewers to work faster.
Choose the test group from sessions that use the same tool and document type as the baseline. If the change also alters the tool, the result cannot separate the effect of the change from the effect of the tool. Record the start date and the first session that used the new process. Keep the baseline sessions in the comparison.
Comparing document types
Different document types carry different reading and coding demands, and pooling them can hide the real pattern. A contract may take longer to read than a short email but require less exception handling. Report the count and the median for each type, and label small groups as descriptive only. Put the count beside the median before comparing one type with another.
If one type dominates the time, check whether the coding guide gives enough examples for that type. A type that looks slow may simply be the type that carries the most judgment. Compare the process, not the people. The next study can test a type-specific guide before applying it to the whole review.
What the evidence does not show
The study does not establish that a faster stage is a better stage. It also does not measure whether a document was coded correctly or whether a privilege call was right. Those are quality questions that need a separate sample. Time is one input to planning, not a verdict on the review.
Do not use the result to promise a client that a review will finish by a certain date or that every document will be read in a certain way. Do not publish document content or reviewer identities. Keep the sample size, the stage definitions, and the review decisions with the result. If the firm needs a legal conclusion, it should ask counsel rather than rely on a time record.
Do not publish reviewer names or document content in the result. Use session codes and document-type labels. If a group is small enough that a reader could identify a person, report the group as a whole or omit it. The purpose is a process improvement, not an individual assessment.
Using the result in a client conversation
A client may ask how the firm plans a document review, and the study supports a careful answer. Describe the window, the number of sessions, the stage distribution, and the support change that was tested. Do not say that the review will be faster or cheaper in every matter. Explain that the measure covers observed work and that the lawyer reviews the substantive calls.
If the client asks for a commitment, offer a process commitment such as a document index, a coding guide, and a named exception owner. Record the commitment and the measure that will show whether it was met. Keep client-specific details out of any public write-up. The study is an internal planning tool, not a claim about the outcome of any matter.
Sources and method
Open each source to confirm its scope, publication date, and applicability before relying on it for a legal workflow decision.