Research question and scope

This protocol measures how often a submitted chargeback response is missing at least one required evidence item. The unit is one submitted packet tied to one disputed transaction. Completeness means every field the network and the client require is present and readable. The measure describes packet quality at submission, not whether the network decided in the merchant's favor.

Set the scope to one merchant account, one card network, and a fixed period. Record the response template version and the evidence checklist in force. If the checklist changed during the period, split the sample at the effective date. A different checklist creates a different denominator.

Keep completeness separate from outcome. A complete packet can still lose because the underlying transaction was invalid. An incomplete packet can win because the customer withdraws the dispute. The purpose is to test the desk's evidence discipline, not to predict the network decision.

State the decision the result will support before collection begins. The client may use it to add a field to intake, change the review step, or shorten the internal deadline. A measure with no decision becomes another monthly report. Write the intended action beside the scope.

Population and unit of analysis

Define the population as every chargeback response submitted during the period. Include packets submitted through a portal, by email, or by a payment provider. The unit is the packet, not the order or the customer. One order with two disputes is two packets when each has its own response.

Exclude drafts that were never submitted and packets withdrawn before review. Record the excluded count and the reason. Mixing drafts into a completeness measure makes the desk look worse for work the network never saw. Report the exclusion so the reader can judge the mix.

Include packets assembled by any team member or tool. Record the queue, the sender role, and whether a template was used. Leaving out an automated assembly path hides part of the evidence process. State the inclusion rule before drawing the sample.

Decide how to treat a packet that was corrected after submission. Score the version the network received first, then note the later correction separately. Record the correction time and the changed field. A repaired packet is still evidence of a first-submission miss.

Definitions that make the measure reproducible

Define a required item as any field the network lists for the dispute reason plus any field the client's policy requires. Publish the list with the checklist version before scoring. Do not add an item after seeing the sample. The definition should be stable across reviewers.

Define a complete packet as one that contains every required item and can be read without opening another system. Define a partial packet as one missing one or two items. Define a failed packet as one missing three or more items or missing the response deadline. Give one example of each before sampling.

Define readability with concrete tests such as legible amounts, dated exports, and labeled exhibits. A file that exists but cannot be opened is missing for practical purposes. Record the test and the result. The reviewer should not have to interpret a broken attachment as present.

Define a late packet by the network's submission timestamp, not by the desk's internal completion time. Record both times. The gap shows whether the internal cutoff is working. A packet that was complete but late is a different failure from an incomplete packet submitted on time.

Sampling and data collection

Draw the sample from the submission log for the full period. Record the export date, the filter, and the row count. Do not build the sample from packets the client remembers or from disputes that were escalated. A selected sample cannot show the everyday packet.

Capture the packet identifier, the transaction, the dispute reason, the queue, the assembler role, the submission time, and the missing items. Add the checklist version and any correction. Store the packet copy with the report. The packet and the checklist are the core evidence.

Check a small overlap sample with a second reviewer. If the reviewers disagree about whether an exhibit counts as proof of delivery, the definition is too vague. Rewrite the definition and retest before the full sample. Agreement matters more than a large sample.

Record the time spent scoring each packet and the number of systems opened. A measure that requires reconstructing every case will not be repeated. Note whether the submission log exposes the packet directly. Collection cost belongs beside the result.

Measuring completeness

Report the share of packets that are complete, partial, and failed. Show the counts behind each share. The three groups describe different levels of evidence quality and should not be collapsed. A single completeness rate hides the packet that missed only the delivery scan.

Break the result down by dispute reason, queue, and assembler role. A fraud dispute may require different fields from a product-not-received dispute. Compare only groups with the same checklist version. A rate without the group size is hard to interpret.

Report the most common missing item and the second most common. A missing delivery scan may point to a carrier export, while a missing customer contact may point to intake. Record one example of each. The pattern should lead to one specific fix.

Report the share of packets submitted after the internal cutoff. A high late share suggests the queue is starting too close to the deadline. Record the median hours before the network deadline. Lateness is an operational signal even when the packet is complete.

Limitations and next action

The measure cannot show whether the network valued the evidence it received. It also cannot show whether a required field was irrelevant for a particular case. Treat the result as adherence to the published checklist. Note the dispute reasons and queues that were not included.

Give the result to the billing owner and the client contact with one change to test. The change might be a new intake field, a shorter checklist, or an earlier internal cutoff. Record the effective date and repeat the measure with the same definitions. A controlled change is easier to judge.

Review the boundary cases before adding a field. A packet may be complete under the network rule but weak for the client's own policy. Separate the two standards in the report. The goal is a packet that can be assembled and reviewed, not a longer form.

Schedule the next measurement for the same week in the following month. Keep the card network and dispute filters unchanged so the comparison is fair. Record any checklist change in the note. A trend built from different requirements is not a trend.