Formula to estimate conversation backlog and its drain time

The backlog is the difference between the conversations that come in and the ones the team manages to close on the same day. It is not a peak problem: it is a pace problem. If 120 come in and 100 are closed, 20 are left over every day, and those 20 add up to yesterday's 20. By the end of the week there are 140 overdue conversations and nobody knows where they came from. This page is meant to put a number on that delay before it shows up in complaints.

The math is simple and can be done with a spreadsheet. What is not simple is deciding what to do with the result: lower the intake (impossible), increase capacity (costly), or accept the delay and prioritize. The formula does not decide, but it tells you how much room you have before the queue becomes unmanageable.

The daily backlog formula

Daily backlog = conversations that come in − conversations that are closed. If the result is positive, it accumulates. If it is negative, it drains. If it is zero, the system is in equilibrium: everything that comes in goes out, but there is no room for the unexpected.

For the math to be useful, both terms must be clearly defined. A conversation "comes in" when a new contact writes or when an old one opens a topic different from the previous one. A conversation "is closed" when the team no longer needs to reply: it was resolved, it was routed to another channel, or the customer stopped replying and it no longer makes sense to continue. Both definitions must be agreed on before measuring, because otherwise each person counts differently and the number is useless for comparing weeks.

How to measure each variable without expensive tools

You don't need an advanced analytics system to fill in this table. The message inbox and a spreadsheet are enough. The most common mistake is counting messages instead of conversations: a customer who sends six messages in a row is one conversation, not six. If you count messages, the backlog inflates and decisions come out wrong.

• Inbound: count new conversations per day for two weeks. If the business has weekly seasonality, two weeks are enough to see the pattern; if it's seasonal by month, measure a full month.
• Closures: define a single closure tag and ask the team to use it when finishing. Without a tag, closures are estimated, and estimates are always optimistic.
• Accumulated backlog: don't count it by hand. Add up the daily B values in the spreadsheet. The sum is what shows whether the problem is a one-day issue or structural.
• Extra capacity: it's measured in hours, not in willingness. If the team has 3 free hours per day and each conversation takes 6 minutes, the extra capacity is 30 conversations per day.
• Average time per conversation: measure it on a sample of 20 real conversations, stopwatch in hand. The mental average always underestimates.

Scenario table: how much accumulates and how fast it drains

The numbers in the table are examples built with the formula, not market data. They serve to show how the result changes depending on the relationship between inbound and closures. Notice that a small, sustained imbalance generates a large delay: 15 conversations per day is 450 in a month.

Reading the table is straightforward: the accumulated backlog grows linearly with the days, but the drainage days depend on how much extra capacity is obtained. Doubling the extra capacity doesn't reduce the backlog: it halves the time it takes to drain. They are two different problems and they are fixed with different measures.

A restaurant case: Friday, Saturday, and the Monday nobody wants to look at

A restaurant with delivery and dine-in measures two weeks and finds the pattern: from Monday to Thursday, 60 conversations come in per day and 55 are closed; Friday and Saturday, 140 come in and 100 are closed. The daily backlog on weekdays is 5; on Friday and Saturday it is 40 each. In a typical week: 5 × 4 + 40 × 2 = 100 accumulated conversations.

On Monday the team tries to drain those 100, but Monday already has its own 60 incoming and barely closes 55. The real extra capacity is 20 conversations (two hours freed up in the morning, with 6-minute conversations). The drain days are 100 ÷ 20 = 5, but since the backlog renews every Friday, it never reaches zero. The number does not say the team works badly: it says the queue builds up over the weekend and they try to pay it off on Monday, when Monday already has its own load.

The average trap: why the backlog is not seen until it explodes

A backlog of 20 conversations per day seems manageable. The problem is that it is not distributed evenly: it accumulates on the days with the highest incoming and they try to drain it on the slow days. If the team does not have an explicit mechanism to attack the queue on slow days, the delay stays there and grows.

That is why the formula has two uses. The first is diagnosis: knowing whether the system is in balance or not. The second is weekly control: if this week's accumulated backlog is greater than last week's, the problem got worse, even if incoming dropped. A backlog that drops does not always mean the team improved: it can mean conversations are being lost, which is worse.

• Stable and low backlog: the system is healthy. Do not touch anything.
• Backlog growing week after week: there is a structural imbalance between incoming and closures. You need extra capacity or reduce incoming.
• Backlog dropping but with complaints: conversations may be closed without resolving them. Review the definition of closure.
• Backlog dropping because incoming fell: it is not an improvement, it is a demand problem. Do not celebrate it.
• Backlog concentrated on one day of the week: it is not backlog, it is a peak. It is solved with shifts, not with capacity.

Checklist to measure the backlog without fooling yourself

1. 1Define what a conversation is and what a closure is. Write it down and make sure the team uses it the same way every day.
2. 2Measure entries and closures for at least 10 business days before drawing conclusions. A single day says nothing.
3. 3Count conversations, not messages. If the system doesn't allow it, do it by hand on a sample and project it.
4. 4Calculate the daily backlog with the subtraction, not with the feeling. The feeling of "we're fine" is what produces the 300-conversation delays.
5. 5Separate the backlog by type of inquiry. A backlog of 100 price inquiries is different from a backlog of 100 complaints.
6. 6Measure the average time per conversation on a real sample, not on what the team thinks it takes.
7. 7Calculate the drainage days with the real extra capacity, not with the desired one. If there are no free hours, there is no extra capacity.
8. 8Review the accumulated backlog every week. The trend matters more than the number of the day.
9. 9If the backlog grows three weeks in a row, escalate the decision: it won't be fixed with more effort from the same team.
10. 10Keep a record of what was done each time the backlog went down. That is what gets repeated later.

What to do with the result: four possible decisions

When the formula gives a positive and sustained backlog, there are four paths and none is free. Reducing entries means filtering inquiries, and that has a commercial cost. Increasing closures means the team works faster, and that has a ceiling. Adding capacity means adding hours or people, and that has a direct cost. Accepting the delay means prioritizing: handle what has the most value first and leave the rest for later, with the risk of losing it.

In practice, the fourth row is usually the most realistic for a small business. Automating the repetitive doesn't replace the team: it changes where time is spent. A beauty business that receives 40 daily inquiries of "how much is a haircut?" or "do you have an appointment on Saturday?" frees those conversations from the human and leaves the human for what needs judgment. Wando's inbox with AI suggestions works like this: the AI proposes the response, the team approves it with one click and the real responses remain as a reference for the next time. That raises C without adding hours, and over time raises K because repetitive inquiries stop reaching the queue.

The backlog isn't measured once: it's monitored

The formula works for the initial diagnosis, but its real value appears when it becomes a weekly control. A backlog of 200 conversations is not a problem if it drops by 50 per week. A backlog of 30 is a problem if it rises by 10 per week. Direction matters more than size.

For the control to work, the variables must be measured the same way every week. If one week you count messages and the next week conversations, the comparison is useless. The discipline of measuring is more important than the precision of the instrument: an approximate but consistent count detects the trend, which is what is needed to decide.

And there is a limit worth stating: this formula measures conversations, not satisfaction. A low backlog can coexist with angry customers if conversations are closed without being resolved. The number serves to size the workload, not to replace reading what customers are saying.