A dialer can measure almost anything, which is why most dashboards measure everything and mean nothing. The test for a metric worth watching is simple: if it moves, do you know what to do? Five pass that test.

1. Answer rate

The share of dials that a human picks up. It is a property of your numbers and your list, not of your agents, and it is the number that limits everything downstream. When it falls, look at caller ID and list quality before you look at the floor.

2. Conversation rate

Of the calls that were answered, how many became an actual conversation rather than an immediate hang-up. This one is about the first ten seconds the connect delay, the opener, and whether the agent was listening when the line opened. It is the most coachable number on the list.

3. The distribution of talk time, not the average

Average talk time hides everything interesting. A floor averaging four minutes might be having uniformly decent conversations, or it might be having a few long good ones and a great many thirty-second failures. Look at the spread. The short tail is where the coaching is.

4. Minutes per connected conversation

Your efficiency figure, and the one that connects the floor to the invoice. Rising minutes per conversation with flat outcomes means time is going into calls that were never going to convert; falling minutes with flat outcomes usually means agents are getting better at qualifying out early, which is good.

5. Callback volume

How many people ring your numbers back. Almost nobody tracks it, and it is the clearest signal that your outbound is landing as legitimate rather than as noise. A floor that gets called back is a floor with a reputation worth protecting.

What to do with them

• Review them weekly, per agent and for the floor. Daily is noise; monthly is too late to act.

• Change one thing at a time. Two changes in a week teach you nothing about either.

• Compare agents on conversation rate, never on dial count. Dial count rewards the wrong behaviour.