Call operations research
A method for reviewing partial intake records from abandoned calls
A review method for partial intake records left by abandoned calls, covering consent, missing-field patterns, contact rules, and follow-up.

Table of contents
An abandoned call can leave a record that is half a request. The name is there, the reason may be missing, and the consent question may never have been asked. The record is tempting to use because it represents a caller who tried to reach the office. It is also easy to misuse because the caller did not finish the interaction. This review describes a method for reading those records without overstepping.
Key takeaways
- Treat a partial record as incomplete evidence, not a finished request.
- Check the consent state before any follow-up contact.
- Group missing fields by the step where the call stopped.
- Separate a caller who hung up from a caller who was cut off.
- Use the pattern to fix the form, not to blame the caller.
Research question and scope
The review asks which fields are missing in partial intake records, where in the process the call stopped, and whether the record supports a follow-up. The unit is the partial record, not the call. A record may contain several missing fields and still support a safe callback. Each record is read once and coded against the same field list.
The scope covers inbound calls that ended before the intake was complete. It includes hang-ups, dropped calls, and calls transferred to voicemail before the form was finished. It excludes calls that were completed and later edited. The review does not evaluate the quality of the service provided before the call ended.
Record the form version, the queue, and the channel. A web form and a phone script may capture different fields. A form change during the period can alter the missing-field pattern without changing caller behavior. The reader should know which version produced the result.
Definitions and unit of analysis
Define a partial record as an intake with at least one required field missing and no completed submission. A record with all fields but no confirmation is also partial. Define a required field as one the office needs before it can act on the request. The definition should come from the office policy, not from the form alone.
Define the stop point as the last step the caller reached before the record ended. The stop point may be a page, a question, or a transfer. Record it as a short label so the pattern can be grouped. A stop point is more useful than a timestamp alone.
Define consent as the caller's explicit agreement to be contacted about the request. A pre-checked box is not the same as an unchecked box that the caller selected. Record the consent state and the evidence, such as a checkbox, a spoken yes, or a written note. If consent is unknown, treat it as absent for follow-up.
Define a safe follow-up as one that uses the channel and time the caller allowed. A phone call to a number left in a partial form may be safe only if the caller agreed to calls. A text may require a separate agreement. The rule should be written before the review begins.
Sampling and consent
Select records from a defined period and a defined queue. Include every partial record if the volume is small, or draw a random sample if it is large. Keep the export date and the filter in the report. A sample of records with phone numbers will overrepresent callers who got further in the form.
Check the consent state before reading the record in detail. If the record contains health or financial details, limit the review to the fields needed for the question. Do not copy sensitive details into the audit file. Use a record identifier instead.
Decide in advance whether the review will lead to follow-up calls. If it will, use the same consent rule as the live team. If it will not, say so in the report and do not contact anyone. A research review is not a reason to override the caller's choice.
Coding missing fields
Code each required field as present, missing, or unclear. A field with a partial value, such as a first name only, is unclear rather than present. Record the stop point beside the field pattern. The combination shows where the form loses people.
Group the records by stop point and by missing-field pattern. A cluster at the phone number field may mean the form asks for it too early. A cluster at the consent question may mean the wording is confusing. The grouping turns a list of gaps into a design question.
Record whether the caller hung up or the call was cut off. A hang-up is a choice, while a dropped call is a technical event. The two have different fixes. Do not merge them into a single abandonment category.
Add a free-text note for anything the codes cannot capture. Keep it short and free of sensitive details. A note such as "caller asked whether the call was recorded" is useful. A note with the caller's full complaint is not needed for this review.
Measuring completion and follow-up
Report the number of partial records and the share of all intake attempts. Show the count beside the share because a small sample can produce a large percentage. Group the records by stop point and by missing fields. The grouping is the main result.
Report the share of records with clear consent for follow-up. Then report the share that received a follow-up within the stated window. A low follow-up rate may reflect a consent problem or a capacity problem. Separate the two before drawing a conclusion.
Measure the outcome of the follow-up when it happens. Record whether the caller completed the request, declined, or did not answer. A completed request after a safe follow-up is a good result. A second abandonment is a signal that the form or the callback timing needs work. The return-call script offers wording for the call.
Measure the time between the partial record and the follow-up. A call after several days may reach a caller who has already solved the problem elsewhere. A call within the window stated on the form is easier to explain. Report the distribution rather than only the average.
Analysis and limitations
The method cannot show why the caller left. The record may end at a field that is unrelated to the reason. The stop point is a clue, not a cause. State that limitation in the report.
The sample may miss callers who never started the form. A caller who heard the greeting and hung up leaves no record. Pair the partial records with the total call volume and the abandonment count. A low partial-record rate may mean a short form or an early exit.
The consent field may be unreliable. A caller may have agreed orally and the agent did not record it. A box may be pre-checked by mistake. Report the evidence beside the state and treat unknown as absent. The conservative rule protects the caller and the office.
The review itself may create a follow-up bias. Agents who know the records are being reviewed may call more often or fill fields differently. Keep the review quiet until the sample is complete. Note the method in the report.
Repeating the review
Repeat the review after a form change with the same definitions and the same window. A change in either can move the result without a change in behavior. Record the form version and the sample dates beside the numbers. The comparison should be reproducible.
Use the result to change one step, such as moving the phone number field later or rewording the consent question. Test the change for two weeks and compare the completion rate and the safe follow-up rate. Keep the old form available in case the new one performs worse. Share the finding with the agents and ask where callers usually stop.
The review should end with a safer record and a clearer form. A partial record is not a failed caller. It is evidence about the step that did not work. Read it carefully, fix one step, and check again.