Every business that moves from mobile phones to a call centre system gets the same unpleasant surprise in week one.
They discover how many people were trying to reach them and giving up.
Why you cannot see them now
On a mobile phone, a caller who gets an engaged tone and rings off leaves no trace you will ever look at. They are not a missed call — the phone was answered, by somebody else. They are simply nothing.
So the number of customers who tried to reach you today is, right now, unknowable. The first week of real reporting usually reveals somewhere between five and fifteen percent of attempted calls ending that way, concentrated in one or two hours.
For a business taking 120 calls a day, 10% is twelve people a day. Sixty a week. That is the number that tends to end the discussion about whether a system is worth KSh 1,780 a month.
What it costs
Three things, in increasing order of expense.
The immediate sale. Somebody who rang to buy and could not get through. Occasionally they ring back. Often they ring somebody else.
The support call that becomes a complaint. A customer with a problem who cannot reach you does not forget the problem. They escalate it — publicly, sometimes.
The habit. This is the real cost. After two or three failed attempts, people stop trying. They have learned that your line does not get answered. You never see that decision, and it is very hard to reverse.
When they happen
Abandoned calls are almost never spread evenly. They cluster, and the clusters are predictable once you can see them.
The mid-morning peak. Most Kenyan desks are busiest between ten and noon. If you are short by one person, that is where it shows.
Lunch. The hour when half the team is away and the callers are not.
Monday. Whatever happened at the weekend arrives at once.
Month end. For SACCOs, lenders and anyone whose customers are paid monthly, this is a different scale of problem. There is a separate piece on it.
The point of measuring is that these are specific, repeating hours. You do not need more staff; you need staff at ten on a Monday.
The three things that actually reduce it
A queue that rings more than one person. This alone fixes most of it. An engaged tone has no patience; a queue does.
A callback offer instead of a wait. The caller keeps their place and hangs up; the system rings them when an agent is free. Most people take it when the wait is over a minute, and it converts an abandoned call into a completed one.
Removing the call entirely. The commonest questions do not need a person. A menu that handles balance enquiries, or an SMS that confirms dispatch before the customer rings to ask, takes work out of the peak permanently. Cheaper than hiring, and it works at month end too.
What not to do
Do not just extend the ringing. Making people wait longer before giving up improves the number without improving anything real.
Do not hide the queue position. Telling someone they are fourth in line respects them. Playing music that implies imminent answer for four minutes does not.
Do not staff for the worst hour of the year. You will carry idle people for the rest of it. Staff for a normal peak and have a plan — overtime, callbacks, a borrowed colleague — for the exceptions.
Measure it first
Before you change anything, get a month of honest numbers. You need to know which hour, which day, and how many.
That is the one thing a mobile phone can never tell you, and the reason the first month with a real system is usually more interesting than anybody expects.