An estate records request log is more than a list of emails. It is the memory of what the team asked, who was asked, which date range and search terms were used, what was produced, and what remains unknown. Without that memory, a later reviewer may mistake silence for a negative result, duplicate an unanswered request, or assume that a custodian searched systems that were never named. A well-designed log makes the research reproducible without pretending that every request resolves the underlying legal question.
This article presents a practical structure for estate professionals. It does not give discovery advice, determine privilege, or prescribe a retention policy. The Bank accounts and services and the Bank accounts and services were checked as examples of authoritative sources with different custodians and purposes.
Give every request an identity
Assign a request number that does not change when the request is revised. Capture the estate matter, requester, custodian, channel, date sent, date range, asset or issue, and requested output. “Please send all records” is hard to audit. “Please search account ending 4482 for statements from January 1, 2022 through the date of death and identify any transfer or beneficiary claim” gives the custodian a defined task.
Keep the request number separate from a production number. One request can produce several files or a staged response. One production can answer multiple rows in the inventory. This distinction preserves provenance when a document is copied into more than one workstream.
Separate request, response, and interpretation
Use three linked records. The request states what was asked. The response records what the custodian said or delivered. The interpretation explains which estate proposition the response supports, contradicts, or leaves open. Do not put all three in a single free-text note. A concise response such as “no responsive records” can mean no records existed, no records were found under the search performed, or the custodian misunderstood the date range.
Quote the response and identify the person, office, or automated portal that supplied it. Save attachments in their original format. Record whether the production is complete, partial, redacted, image-only, or subject to a stated limitation. If a response is verbal, memorialize the date, speaker, and exact scope as soon as possible.
Worked example: a bank request with a vague result
Suppose the inventory lists a bank account ending 4482, but the initial request asks the bank for “all estate accounts.” The bank replies that it found no accounts. That response is not enough to close the row. The log should show the original wording, the date, the search identifier supplied, the bank’s search boundary if stated, and the limitation that the known account fragment was not included.
Send a narrowed follow-up identifying the account fragment, known names, relevant date range, and requested statement periods. If the bank then confirms that the account was closed before death, save the closure date and supporting statement. If it says that it cannot search without a legal representative’s authorization, classify the request as blocked by authorization rather than as a negative result. The log preserves both the work performed and the reason a next step is needed.
Define search scope before sending
Every request should answer: who or what is being searched, which systems or record series are included, what dates apply, which identifiers are used, what format is requested, and what the custodian should say if the search is incomplete. A county recorder request might name grantor and grantee variants, parcel number, instrument type, and historical years. A court request might name the case number, docket, pleadings, orders, and attachments.
Do not ask a custodian to make a legal conclusion. Ask for documents, fields, or search results. “Confirm the estate owns the parcel” should become “provide recorded instruments in the named date range and identify the parcel description used.” The professional can then evaluate the legal effect.
Track follow-ups without erasing history
When a request changes, retain the original and create a revision or linked follow-up. Do not edit the first request until it appears to say what was actually sent. Record reminders, escalation, bounced messages, portal errors, fee notices, and returned mail. A request can be operationally complete even when it is substantively insufficient; use separate status fields for those conditions.
Useful statuses include drafted, sent, acknowledged, partial production, complete production, no responsive records, unable to search, awaiting authorization, disputed, and closed with professional review. Define the status in a data dictionary so different people do not use “closed” to mean different things.
Preserve negative evidence carefully
The phrase “no records found” must be tied to the custodian’s stated scope. Record the search terms, variants, date range, role filters, result count, and whether the source was an index or underlying file. If the custodian cannot disclose its search method, state that limitation. A negative result may be a lead for another custodian, a request for a certified search, or a professional judgment that the remaining uncertainty is acceptable.
Do not convert an unanswered request into a negative result. Do not treat a bounced email as a request received. Do not infer that a portal’s empty screen means the historical archive was searched. Each condition belongs in the log.
Connect requests to the evidence map
Link a request to the inventory row, parcel crosswalk, person identity table, tax question, or docket event it is intended to test. A request for a deed may support identity, ownership, legal description, and date questions, but the same document may not resolve all four. Let each linked proposition carry its own conclusion.
When production arrives, record file names, page counts, hashes where required, confidentiality classification, and source location. Preserve the original and create a working copy for redaction or annotation. If a source is public, record its URL and retrieval date. If it is private, record the custodian and authorization basis without exposing private information in a public article.
When a request produces mixed results
A custodian may deliver statements but omit correspondence, provide an index but not images, or answer one date range while leaving another unanswered. Split the response into production rows instead of marking the whole request complete. Identify the documents received, the period covered, and the gap. A partial production can be highly useful without being complete.
If two custodians answer the same question differently, preserve both replies and compare their definitions. One may search current accounts while the other searches closed accounts; one may use a legal name while the other uses a customer number. The resolution is a follow-up question about scope, not an editorial decision that one response is more convenient.
Date that comparison and link it to exact request numbers so later reviewers can reconstruct which answer came first.
Quality review and handoff
A second reviewer should select one completed request, one partial response, and one negative result. They should be able to find the original request, response, production, interpretation, and next action. Correct broken links, ambiguous statuses, missing date types, and rows that claim completion without a response.
Close a request without erasing uncertainty
A request can be operationally closed while the underlying proposition remains unresolved. Record the response, production index, search limitation, and professional next action separately. If the custodian produced only a current statement, do not mark historical coverage complete. If the response says no records were found but gives no search scope, preserve that wording and ask whether a narrower request or authorization would produce a different result.
Make the log answer the next reviewer’s question
The strongest request log is not the longest one; it is the one that lets a new reviewer reproduce the search boundary. For each request, preserve the identifier, custodian, date range, names and parcel identifiers supplied, delivery method, response date, production list, and unresolved proposition. In a mixed response, separate a complete answer about one account from an unanswered question about a second account. That distinction changes the next action: a targeted follow-up may be appropriate, while a broad duplicate request wastes time and can create conflicting records. The log should therefore explain both what the custodian answered and what the estate still does not know.
Frequently asked questions
Is a request log the same as a document index?
No. The request log records the search activity and responses; the production index describes the files received.
Should a custodian’s “no records” response close the issue?
Only after its scope is understood and the responsible professional decides whether the remaining uncertainty matters.
Can a researcher ask for a legal conclusion?
The request should seek documents and search results. Counsel or another professional should draw the legal conclusion.
What makes a log defensible?
Stable identifiers, exact requests, dated responses, preserved limitations, linked productions, and visible next actions.