How Many Human Attention Hours Your WhatsApp Requires: Formula and Table by Time Slot

What this formula is for (and what it is not)

This page does not tell you how many people to hire. It gives you a formula to go from a number you already have —messages per month— to a number you do not have —hours of human work needed— and a table to distribute those hours by time slot. With that you can decide whether you are short on staff, whether you need to change your service hours, or whether the bottleneck is somewhere else.

The calculation is done with your data, not ours. The only thing we provide is the formulas and the conversion factors you need to choose. If you do not have the data, the audit section tells you which data to gather before doing the calculation: running the formula with made-up numbers is worse than not running it at all.

The base formula, in one line

Human hours per month = (Incoming messages per month × Human attention factor × Average resolution time) ÷ 60.

The three terms are estimated differently and it is best not to mix them. Incoming messages per month comes from your inbox or your WhatsApp report. Human attention factor is the proportion of messages that actually require a person to write, review or decide something: those resolved with a template, an automatic greeting or a fixed piece of information do not count. Average resolution time is how many minutes of human work each of those messages consumes, counting reading the history, writing, waiting and coming back.

The ÷60 is only to convert minutes to hours. If your average time is in seconds, divide by 3600.

In the example, 3,000 messages per month turn into 120 hours of human work. That number alone says nothing: you have to compare it against the hours your team actually has available to handle WhatsApp. If you have two people each working half-time, that's about 160 monthly hours of gross capacity, and you already know you're at the limit before counting vacations, holidays and absences.

How to estimate the human attention factor without making it up

The human attention factor is the number people get wrong most often, because they estimate it by feel. The method that works is to sample real conversations and classify them into three buckets: automatic (resolved without anyone writing), assisted (a person reviewed and approved or corrected an already drafted text) and manual (a person wrote from scratch or made a decision).

The human attention factor is the sum of assisted plus manual, divided by the total sample. Assisted ones count, but they consume fewer minutes than manual ones: that's why the average resolution time has to come from timing each bucket separately and then averaging with weights.

1. 1Pick 100 conversations from the last two weeks, at random and from different days of the week.
2. 2Classify each one as automatic, assisted or manual, with criteria written down before you start.
3. 3Time 20 conversations from each bucket that has at least 20 cases; if a bucket has fewer, time all of them.
4. 4Calculate the average time per bucket and the weighted average by the proportion of each bucket.
5. 5Repeat the sampling every three months or every time you change the service flow: the factor moves.

A common mistake is counting conversations that started with an automated greeting but later needed a person as automated. Classification is done by how the conversation ended, not by how it started.

Conversion table by time slot

The monthly total spread evenly is a fiction. WhatsApp support has a marked daily curve: there are time slots where most of the volume comes in and time slots where almost nothing comes in. This table converts a monthly volume into hours per time slot, applying a typical retail distribution. Replace the percentages with your own before using it to make decisions.

The column that matters is the last one: it tells you how many human hours fall in each time slot. If 22% of the volume falls between 10:00 and 12:00 and your team starts at 10:00, that time slot is handled well. If your team starts at 12:00, the morning peak is answered late and the afternoon peak is answered with the morning backlog on top of it.

A detail that changes the calculation: the volume of a time slot is not the workload of that time slot. A message that comes in at 23:00 is answered at 9:00 the next day, so its minute of work is consumed in the following time slot. To size shifts, the workload shifts to the time slot when people are actually working.

From hours to people: the step almost everyone gets wrong

Going from hours to people is not dividing by 8. A person does not handle WhatsApp for 8 hours straight: they alternate with other tasks, take breaks, get sick and take vacations. The reasonable conversion uses effective hours per workday, which in customer service are usually quite a bit less than presence hours, and an absenteeism factor.

With 120 human hours needed per month and a capacity of 120 hours per person, the result is exactly one person. Exactly one person does not exist: any absence leaves the inbox unattended. That is why the number you are looking for is not the minimum, it is the minimum plus a margin that covers the peak and absences. The right question is not how many people I need, but how many hours of capacity I need to cover the peak of the worst week of the month.

What changes in the calculation when a tool with suggestions comes in

An inbox with AI suggestions that learns from the business's real replies does not eliminate human hours: it reduces them and moves them from bucket to bucket. Simple manual conversations become assisted, and assisted ones drop from minutes because the text is already written and you only have to approve or correct it. The human service factor does not drop to zero, because approval is still human.

To estimate the effect, do not change the factor: change the distribution between buckets and run the formula again. If before, 60% of conversations were simple manual and after adopting suggestions they become assisted, the weighted average time drops and the monthly hours drop with it. That is the calculation worth doing before deciding, not after.

Wando works like this: the AI suggests replies and the team approves them with one click, and it learns from the business's real replies instead of requiring programmed flows. If you want to see how that lands in your inbox, you can try it without a card, but you still have to do the hours calculation with your own numbers.

Audit: what to check before trusting the result

• Does the incoming message count exclude the ones your own team sends? Adding them inflates the volume and ruins the calculation.
• Does the count come from the actual message date and not the date it was loaded into the database? If there was a history import, the distribution by time can end up concentrated in a couple of days and be useless.
• Does the sample of 100 conversations cover weekdays and weekends, morning and afternoon? A sample from a single shift does not represent the month.
• Does the average resolution time include the waiting time between messages? If it doesn't, it underestimates the load.
• Is the capacity per person in effective hours or in presence hours? Mixing them gives a number that doesn't hold up.
• Is the peak of the worst week of the month covered? Averaging the month hides the day the inbox goes down.
• Was the distribution by time slot measured with the correct time zone? A three-hour shift changes the shift that needs reinforcement.
• Is the calculation redone after every flow change? A human attention factor from six months ago no longer describes what happens today.

Mistakes that make the formula give you anything

The most common mistake is using the total number of messages as if they all required human work. In an inbox with templates and fixed replies, a significant part of the volume is resolved without anyone writing, and counting it inflates the need for people. The second mistake is timing the resolution time with the clock of the team's best agent: the average has to come from different people, including the newest one.

The third mistake is dividing by 8 hours and by 30 days. That calculation gives a theoretical capacity that doesn't exist: nobody works 30 days, nor 8 effective hours, nor without absences. The fourth is not separating volume from load: messages that come in outside business hours consume minutes from the next time slot and have to be assigned there.

And the fifth, the most expensive one: doing the calculation once and never doing it again. Volume changes by season, the human attention factor changes when templates are added, and capacity changes when someone resigns. The formula works if it's run again, not if it's filed away.