Document review budgets: distinguish email families from reviewable records

An email count can understate a document review project when attachments are separate reviewable records. A duplicate-file count can also mislead if the approved workflow preserves the same attachment in different message families. Build the workload from the actual review population rather than multiplying a collection headline by a generic rate.

Relativity distinguishes item-level duplicate identification from family-level treatment, where a message and its attachments are considered together. Its documentation shows why the same attached file can remain associated with different email families. That distinction affects counting, but the responsible legal or investigation team must determine the approved treatment for the matter.

This is a budgeting method. It does not determine relevance, privilege, preservation, disclosure or which records may be excluded. Keep those decisions with the authorized specialists and preserve the required source and relationship information while estimating the work their instructions create.

Suppose a hypothetical processed collection contains 12,000 parent email messages and 6,000 attachment records, totaling 18,000 records. Under the approved fictional workflow, exact duplicate families account for 4,000 parent messages and 3,000 attached records that are excluded from the active review population while required evidence is preserved.

The remaining population contains 8,000 parent messages and 3,000 attachment records, or 11,000 reviewable records. A budget based only on the 8,000 parent messages would omit the attachments. A budget based on all 18,000 processed records would ignore the approved reduction already established for the active review.

Assume 1,000 of the remaining attachment records share content with attachments in other retained families. In this example, the approved review approach retains those family relationships and their reviewable records. Do not subtract another 1,000 automatically merely because a content-level duplicate report can identify them.

Suppose a representative pilot supports an illustrative planning rate of fifty-five reviewable records per hour for this defined population and task. Reviewing 11,000 records requires 200 hours. At an invented reviewer cost of $65 per hour, the primary review estimate is $13,000.

Using only 8,000 parent messages with that same record-based rate would estimate approximately 145.45 hours and $9,454.55. It would understate the defined primary review cost by approximately $3,545.45. The problem is a mismatch between the counted population and the throughput unit, not an arithmetic error in division.

A family-based productivity measure would require its own pilot and definition. One family can contain a short message and no attachment, while another includes several lengthy files. Do not rename a records-per-hour rate as families per hour to make the smaller count appear compatible.

Assume the approved plan also includes twenty hours of a separate quality-review task at a fictional $80 per hour, costing $1,600. This work is outside the primary review rate in the example. If a real quote already includes the same task, adding it again would overstate the budget.

Suppose the provider confirms eighty billable hosting gigabytes for three months at an invented $10 per gigabyte-month. Hosting costs $2,400. Add a defined $900 setup charge, and the illustrated project scope totals $17,900: $13,000 of primary review, $1,600 of quality work, $2,400 of hosting and $900 of setup.

The eighty-gigabyte amount is an independent assumed billing measurement, not a conversion from 11,000 documents. File counts and data volume have no fixed relationship. Confirm whether the hosting meter uses the relevant native, expanded, stored or other defined volume and how the provider measures the billed period.

The fifty-five-record rate is an example assumption, not an industry benchmark. Establish a rate from a representative sample of the actual review tasks, document types and required decisions. A short email and a large spreadsheet can require different effort even when each contributes one record to the population.

Record whether the pilot includes time spent on required coding, escalation and ordinary support. Keep separately budgeted activities identifiable so the model neither omits them nor counts them twice. The team needs a clear explanation of what a productive review hour includes.

Update the forecast if the remaining material differs from the pilot. A later collection containing unfamiliar formats, language requirements or more complex review instructions can change workload without changing the nominal record count. Obtain the authorized scope and a supported revised estimate rather than assuming the original average must hold.

Maintain the active review count by release, with the approved reason for additions, exclusions and re-review. A later processing wave can overlap an earlier collection, but its treatment depends on the established method and matter requirements. A new import identifier is not sufficient evidence that every record creates wholly new review work.

Preserve the distinction between work completed and records still active. Re-review after an instruction change is an additional task even if the underlying document count is unchanged. Conversely, excluding an unreviewed duplicate under an approved process can reduce remaining work without reversing hours already spent.

Reconcile billed hours or units to the provider's agreed activity records. For finance, aggregate counts and appropriate matter references may be enough; detailed document content should remain within the authorized review environment. Do not circulate sensitive material merely to explain a difference in a budget total.

Show the approved record population, pilot throughput, hourly scope, quality work, billable hosting volume and expected duration. An extension can increase hosting expense without increasing review records, while a new collection can increase records without changing the hosting rate. Each change should have a visible driver.

The resulting budget gives the matter owner a clear view of the cost implied by the approved workflow. It also makes a provider's invoice easier to assess because parent messages, attachments, reviewable records and hosting units have each retained their own meaning.