Free tool · Due Diligence
How Many Documents Must You Review in Due Diligence? A Sampling Planner
You rarely have to read every document in a data room, and in most deals you cannot. The usual method is to divide the documents into groups by risk, read the high-risk groups in full and draw a random sample from the rest. If a random sample of 60 documents from a large group shows no problem, then at about 95 per cent confidence no more than about 5 per cent of the group has it (the rule of three: roughly 3 divided by the sample size). The ceiling is larger for a small group, and zero if you read every document. A sample picked by hand, for example the first documents in a folder, supports no statement about the rest. When a sample does find a problem, widen it or read the whole group for that issue. The report should state what was read, how it was chosen and what that allows you to say.
Tool · Sampling Planner
120 of 1288 documents to review (9%), within a capacity of 120.
| Group | Risk | Documents | To review | Coverage | Ceiling if none found |
|---|---|---|---|---|---|
| Top 10 customer contracts High risk: review every document | high | 10 | 10 | 100% | none: all reviewed |
| Supplier contracts Medium risk: sample for a bound on the defect rate | medium | 120 | 62 | 52% | up to 3% (4 documents) |
| Employment contracts Low risk: a smaller sample | low | 340 | 15 | 4% | up to 18% (60 documents) |
| Leases and property documents High risk: review every document | high | 18 | 18 | 100% | none: all reviewed |
| NDAs and routine correspondence Low risk: a smaller sample | low | 800 | 15 | 2% | up to 18% (143 documents) |
- For each sampled group, the last column shows the highest share of problem documents that is still consistent (at 95%) with finding none in the sample. It is a ceiling, not an estimate, and holds only if the sample was drawn at random.
- A random sample tells you about the rate of problems, not whether a particular document is clean. Review any document that a red flag points to, whatever its group.
- If the sample finds a problem, stop treating the group as low risk: widen the sample or review the group in full.
A data room can hold thousands of documents and the team cannot read them all. Enter the groups of documents, how many there are and how risky each group is, and your review capacity (how many documents the team can read). The planner reads every high-risk document, gives each other group a useful sample and shows the highest share of problem documents that is still consistent (at 95%) with finding none. The figures are for random samples: documents picked because they looked interesting do not support a bound. Invented data. Runs in your browser; nothing is stored or sent.
How to use it
- Enter each group of documents with its number and risk level.
- Enter how many documents your team can review.
- Read which groups are reviewed in full and the size of each sample.
- Read the ceiling for each group if the sample is clean, and the warnings when capacity is short.
Common questions
What is the rule of three?
If you find no problem in a random sample of n documents from a large group, then at about 95 per cent confidence the rate of problem documents is at most about 3 divided by n.
Is a clean sample proof that there are no problems?
No. It gives a ceiling on how common the problem could be. The true rate may be zero, but the sample cannot show that.
Why must the sample be random?
Only a random sample supports statements about the rest of the group. A hand-picked sample is a good way to find problems and a poor way to find out how common they are.
Learn the law behind the tool
This tool is part of our Due Diligence course, with lectures, concept maps, flashcards, interactive exercises and a capstone case. Due Diligence and Data Rooms: Sampling and Coverage.
More free law tools
General information, simplified, mainly on American law, not legal advice. States and countries differ, and real cases turn on facts that a tool cannot capture.