Method note: this brief proposes a source-led review approach. It does not claim that private client or firm data has been collected.
Question and scope
This protocol asks how much invoice review time goes to duplicate, split, and unsupported cost lines. The unit is a review minute attached to a cost line, and the measure is the active time spent identifying and resolving the issue before the invoice is released. The study covers the cost review stage, not the vendor payment process. It does not attempt to estimate the total cost of the billing function.
Scope the study to invoices with at least one cost line reviewed during the period. Include invoices that were released, held, and adjusted so the denominator reflects the full workload. Exclude invoices with no cost lines, and record the number excluded. State the billing system, the cost ledger, and the person who owns the time log.
Defining the event
Define a duplicate as a cost line that repeats an amount already posted for the same matter, vendor, and period. Define a split as a single vendor charge divided across invoices or lines in a way that obscures the original transaction. Define unsupported as a cost with no receipt, no matter reference, or no client authorization. Write two examples and one boundary case so reviewers apply the same rule.
Distinguish a legitimate multiple charge from a duplicate. Two court reporter invoices for two separate hearings are not duplicates even when the amount is identical. A monthly courier summary that lists several deliveries is not a split merely because it has several lines. Code the evidence that shows the charge is separate, not just the similarity of the amount.
Population and denominator
The population is every cost line reviewed in the period, counted once per invoice. Report the number of invoices, the number of cost lines, the number excluded, and the number in the final sample. If a line was reviewed more than once, count the active minutes from each pass but keep the line as one unit. State how rework was handled in the coding guide.
Report the review minutes per cost line and the share of minutes spent on flagged lines. A high share can reflect a small number of complex invoices, so report the number of invoices behind the share. Keep the raw time entries with the analysis file. The denominator should include the lines that were cleared quickly, not only the difficult ones.
Evidence sources
Use the cost ledger, the vendor invoice, the client invoice, and the reviewer's time log as the primary evidence. Compare the line with the ledger and the original invoice, and record the date each was observed. If the time log is missing, mark the line as incomplete instead of estimating from memory. The evidence log should let a second reviewer repeat the match.
Rank the sources so reviewers know what to trust first. A contemporaneous time log is stronger than a later estimate, and a vendor invoice is stronger than a summary line in the billing system. A client authorization is stronger than an internal assumption. Record the evidence type for each line so the analysis can separate a documentation gap from a real review burden.
Sampling
Sample across vendors, cost types, and invoice sizes so one large invoice does not define the result. Include lines that were cleared without a flag, because the review time for a clean line is part of the baseline. Draw the sample before reading the outcome so the selection is not biased toward the difficult cases. If the population is small, review every line and extend the period.
Stratify the sample by whether the cost was received through a vendor statement or a per-matter invoice. A monthly statement may require more allocation work and more duplicate checks. Report the minutes per line for each stratum rather than pooling them. The reader should be able to see where the review burden actually sits.
Coding the record
Use a short code list that points to a specific repair. Code each flagged line as duplicate, split, unsupported, rate variance, or other, and record the stage where it was found. Add an unknown category for lines the evidence cannot resolve. Code the outcome as cleared, adjusted, or escalated so the review time can be interpreted.
Have a second reviewer code a sample of flagged lines and compare the results. Resolve disagreements by returning to the ledger and the vendor invoice, not by majority vote. Report the agreement rate and revise any code that is applied inconsistently. If the reviewers cannot agree on what counts as a duplicate, stop and fix the definition before continuing.
Measuring the variable
Measure active review minutes with a short log completed at the moment of the work. Separate the time spent searching, comparing, deciding, and communicating so the largest component is visible. Report the median and the range rather than a single average, because one complex invoice can dominate a mean. Keep the raw entries with the analysis file so a reviewer can check the calculation.
Do not convert the minutes to a dollar figure without an approved labor rate and a clear statement that it is an internal estimate. The study measures effort, not price. If the firm wants a cost view, run it separately with the finance owner. A precise number built on rounded guesses is less useful than a range with a clear source.
Analysis and reporting
Report the overall minutes per cost line, the share spent on flagged lines, and the number of invoices behind each figure. Show the median and range for each flag type and the most common stage where the issue was found. Compare only groups with similar vendor and cost mixes. Include the clean lines so the report shows the full distribution.
Give the reader one table that answers the main question and put the coding detail in an appendix. Write the finding in plain language before presenting the technical definition. Include the cases that contradict the pattern rather than hiding them in an average. A report that cannot be acted on is just another record of the past.
Limits and sources of error
Self-reported time can be rounded, and a missing log is not proof that no review occurred. Report the completeness of the time evidence before drawing a conclusion. Small samples, changing vendor arrangements, and reviewer disagreement can all limit interpretation. A line that was cleared quickly may still have required a check that the reviewer did not log.
Avoid comparing invoices that differ in cost volume or client requirements. A matter with many small costs will produce a different review pattern from one with a single expert fee. Describe the context beside the minutes and avoid ranking individual reviewers. The purpose is to improve the intake and the queue, not to judge professional worth.
Privacy and retention
Store the invoices, cost ledger, and time logs in a controlled folder with access limited to the study team. Remove client names, amounts, and vendor details from the analysis file when they are not needed. If an example is shared, use a role label and redact the client and the specific amount. Keep the unredacted evidence separate from the working analysis.
Do not circulate a person-level or client-level list about review time outside the management context. Aggregate the results before sharing them more broadly, and check that a small group cannot be re-identified. The same rule applies to screenshots of the billing system and to quoted vendor messages. Privacy is part of the method, not an afterthought added at the end.
Applying the result
Choose one change that matches the most common flag. If duplicates dominate, add a vendor invoice number check at intake. If splits dominate, require the original invoice to be attached before allocation. If unsupported costs dominate, add a receipt and authorization prompt to the cost template. Test the change on one vendor before applying it broadly.
A lower review time is not the only useful outcome. Fewer client questions and a cleaner cost ledger can improve the process even when the minutes stay similar. Report the direct result and the service result separately. Stop the change if it pushes the check into a private channel where it cannot be reviewed.
Review cadence
Repeat the study each invoice cycle and after any change to the vendor mix, the cost template, or the billing system. Keep the definitions and the unit stable so periods can be compared. Report the lines that were excluded or still open beside the main measure. Review the coding guide after the first repetition and keep the changes visible.
Review the result with the billing manager and the cost ledger owner, not only with the reviewer. Ask which fix is easiest to sustain and which flag creates the most rework. Set one change and a follow-up date, then check whether the change held. The next study should test the repair, not just repeat the same count.
Put the guidance to work
Choose one recurring billing workflow, identify its authoritative source records, and name the owner who reviews exceptions. We can help turn that boundary into a focused support role.
Discuss your workflow


