An inbound call center answers calls somebody else decided to make. An outbound one makes calls somebody else did not ask for. That is the whole distinction, and nearly everything else follows from it.
Most teams end up doing both. That is fine, and usually correct. What is not fine is doing both without noticing, because the two pull in opposite directions and the tension shows up as a team that is bad at each.
Inbound: you do not control the arrivals
The defining feature of inbound work is that the volume happens to you. Your customers decide when to ring — at month end, after a power cut, the morning your advert runs — and your only lever is whether somebody is there.
That makes inbound a scheduling problem. The question is never “how many calls can we make today” but “is the right number of people on the phone during the hour the calls arrive”. Get that wrong by two people in the peak hour and the queue does not grow by two calls; it grows by far more, because queues behave badly near capacity.
What you measure follows. Service level — the share of calls answered inside a target, say 80% within 30 seconds. Abandonment, which is the people who gave up, and the only inbound number with revenue directly attached. Average handling time, watched for drift rather than driven down. First contact resolution, because a call that has to happen twice cost you twice.
The staffing is the arithmetic in the piece on starting a call center: peak-hour calls, times handling time, plus spare capacity, divided by real availability. The spare capacity is the part people cut first and should not.
Outbound: you control everything except the answer
Outbound work is a throughput problem. You decide the volume, so the questions are how many attempts an agent can make an hour, what share connect to a human, and what share of those produce the outcome you wanted.
The measures are therefore completely different. Contact rate — connections per attempt, which depends enormously on the hour of day and the number you are calling from. Conversion per contact. Attempts per agent hour. Cost per acquisition, which is the only one the business outside the call center cares about.
Two practical things dominate outbound in Kenya. The first is the number you call from: a recognisable local number gets answered and a strange one does not, and this single variable moves contact rate more than any script. The second is timing — a list worked at 10am and the same list worked at 6pm are not the same list, and most teams discover this by accident months in.
There is also a legal floor. Calling people who did not ask to be called means consent, a suppression list that is actually honoured, and a record of where each number came from. The Data Protection Act applies to your calling list exactly as it applies to your recordings.
Why the two do not mix easily
An agent on an outbound list is productive when they are dialling. An agent waiting for an inbound call is productive when they are available — which looks, from across the room, like doing nothing. Put both jobs on one person and the inbound service level loses, every time, because the outbound work is visibly in front of them and the inbound call has not happened yet.
The skills differ too. Inbound is diagnosis under pressure: a stranger with a problem, often annoyed, and you have no idea which problem until they start. Outbound is resilience and routine: the same opening forty times, mostly refused, and the discipline to make the forty-first properly. Good inbound agents are often indifferent at outbound and the reverse is equally true. Hiring for one and assigning the other is how people leave.
Blended, and when it is honest
“Blended” means an agent takes inbound calls and is handed outbound work when the queue is empty. Done properly it is the most efficient arrangement available to a small team, and it is why a six-person desk can do the work of two separate ones.
Done badly it is a way of hiding understaffing. The giveaway is a service level that quietly slips while outbound numbers look healthy, because every agent is on a list when the calls arrive. If you blend, the inbound measure is the one that must hold; the outbound work is what fills the gaps, not the other way around.
What this means for what you buy
The two jobs want different things from call center software, and pricing shape matters more than the feature list.
- Inbound wants queues that ring a team, a menu, callbacks instead of hold, hour-by-hour reporting, and free incoming calls. If your platform charges per inbound minute, every success at getting customers to ring you raises your bill.
- Outbound wants campaign lists, click to call, outcome codes against every number, a sensible caller ID, and per-second billing — because a nine-second “not interested” should cost nine seconds, not a minute.
Both of those are on every plan here, and incoming calls are free; outgoing is KSh 3.00 a minute billed by the second. The rate card is public, and the overview of what the software does covers the rest. There is no predictive dialler yet, which is worth knowing if you are planning a large outbound floor rather than a blended desk.
The question to actually answer
Not “inbound or outbound” in the abstract, but: which of these, done badly, is currently costing us money?
For most Kenyan businesses the answer is inbound, because the customers are already ringing and nobody is counting how many give up. Fix that first. Outbound is a growth project and it will still be there in three months; the abandoned calls from this morning are gone.