Pending Queue Audit Checklist: Unassigned, Untagged and Aged-Out Conversations

What this audit is for

The pending queue is everything that came in and still has no owner, no classification, or has been waiting too long. It is not the same as the inbox of the day: the inbox shows what is happening now, the pending queue shows what is piling up. A queue that grows silently is the first symptom that service has broken down, and it is detected before the customer complains.

This checklist runs on the queue, not on the individual conversation. The idea is that you can tick item by item, pull three numbers (unassigned, untagged, overdue) and only then decide what to do. If the three numbers are at zero, the queue is healthy and there is nothing to fix that day.

Before you start: define the age cutoff

Without a defined cutoff, 'overdue' is an opinion. Before opening the inbox, write three numbers on a piece of paper: how long a new inquiry can wait without going cold, how long an inquiry that has already been answered can wait for the customer's reply, and how long a case forwarded to another area can wait. Those three numbers are your yardstick and they do not change during the audit.

• New inquiry with no first response: define the maximum in minutes or hours according to your operation.
• Conversation already answered waiting on the customer: the clock runs for the customer, not for you.
• Case forwarded to another area: define it in business hours, not in calendar hours.
• Campaign sent with an incoming response: it has its own cutoff, usually shorter.

If you have no idea what number to use, run the audit for a week without a cutoff and look at the real age distribution of your queue. Then set the cutoff where the problem is concentrated, not where you would like it to be.

Block 1: unassigned conversations

An unassigned conversation is one that came in, exists, and nobody took it. It may be unread, or it may have been read by someone who didn't keep it. Both are the same problem: it has no owner. The goal of this block is that no conversation is left without someone responsible at the close of the audit.

• Count how many open conversations have no assigned agent. That is your first number.
• Split the unassigned ones into two groups: those that were never opened and those that were opened and abandoned. The second group is more serious.
• Check whether there are conversations assigned to an agent who is not working today (day off, leave, night shift). An assignment to someone who is absent is an unassigned conversation in disguise.
• Verify that the assignment is explicit and not by default: 'everyone sees everything' is not an assignment.
• Check whether there are conversations assigned to more than one person. Double assignment makes both wait for the other to answer.
• Note how many of the unassigned ones exceed your age cutoff. Those go first.

Block 2: untagged conversations

Tags are what allow you to count reasons, not conversations. An untagged queue cannot be analyzed: you don't know whether volume went up because of price inquiries, complaints, or a campaign. This block does not seek for you to tag everything; it seeks for you to detect what percentage of the queue was left unclassified and why.

• Count the open conversations with no tag. That is your second number.
• Split the untagged ones by age: recent ones may be an oversight, old ones are a process gap.
• Check whether there are tags created that nobody uses. A dead tag confuses more than it organizes.
• Verify that reason tags are mutually exclusive. If a conversation can carry two contradictory reasons, the later count is useless.
• Check whether closed conversations are tagged at closing or remain unclassified forever.
• Note how many conversations carry a reason tag but not a status tag (or the other way around). That mix breaks any report.

A useful shortcut: if more than 20% of your queue is untagged, the problem is not the team's discipline, it's that there are too many tags or they are poorly defined. Fewer tags, better used, deliver more than a huge tree that nobody completes.

Block 3: conversations overdue by age

Here it doesn't matter whether it has an owner or a tag: what matters is how long it has been waiting. An overdue conversation is one that exceeded your threshold without movement. This block is the one that produces the most uncomfortable and the most useful number, because it measures the real damage to the customer.

• Count the open conversations whose last activity exceeds your threshold. That is your third number.
• Separate the overdue ones by direction: waiting for your reply, or waiting for the customer's reply. The first ones are your responsibility, the second ones are not.
• Among those waiting for your reply, separate the ones that never had a first reply from the ones that stalled mid-conversation.
• Flag the overdue ones that are also unassigned. Those are the critical ones: nobody is going to answer them if you don't pick them up in the audit.
• Check whether the overdue ones are concentrated in a time slot. If 80% piled up outside business hours, the problem is coverage, not speed.
• Check whether there are overdue conversations that were already resolved through another channel (phone, front desk). Closing them is part of the audit.

Summary table: what gets reviewed and how to know it's OK

How the three numbers are read together

The three numbers alone say nothing. Read together, they locate the problem at a different stage of the process. If there are many unassigned and few overdue, the bottleneck is in the distribution. If there are many untagged and few unassigned, the problem is the classification criteria. If there are many overdue and few unassigned, the problem is capacity or time coverage.

What to do when something fails

The audit does not end at the count. Each high number has a different action and none of them is 'talk to the team'. These are the corrections that follow from each case.

• If the problem is unassigned: build a distribution rule by shift and by topic, and assign on opening, not on replying. A conversation that comes in without an owner was already born overdue.
• If the problem is untagged: reduce the tag list to the ones you will actually use to decide something. If a tag does not change any action, remove it.
• If the problem is overdue due to coverage: review the hourly distribution of incoming messages before hiring. You may have excess capacity on one shift and a shortage on another.
• If the problem is overdue mid-conversation: the issue is not speed but follow-up. Define a recontact reminder with a deadline.
• If the problem is assigned to absent people: the assignment must fall to someone responsible on the active shift when the owner is not there.
• If the problem is that you cannot measure anything because the data is mixed up: first organize the capture, then audit. Measuring on dirty data gives false conclusions.

Audit frequency and log

An audit run once is useless. What is useful is the series: the same three numbers, measured with the same yardstick, at the same frequency. With three runs you can already see whether the queue is getting organized or whether the problem is moving from one stage to another.

• Always run the audit on the same day and at the same time. If the moment changes, the number changes.
• Write down the three numbers with the date. Without a date, the series does not exist.
• Record who did the audit. If the same person always does it, the bias is also the same.
• Do not change the age threshold in the middle of the series. If you change it, start a new series.
• Review the series every month, not every day. The queue moves slowly and daily noise is misleading.

If your team handles WhatsApp with a tool that suggests replies and approves them with one click, like Wando, the audit runs on the same inbox where the team works: the three numbers come from there. What does not change is the criterion: the tool does not decide what an overdue conversation is, the business defines that.