A calling dashboard will show you a dozen numbers. Three of them change what you do next; the rest are reassurance. Knowing which is which saves a lot of staring.
The three that matter
Connect rate
What share of dials reached a person. This is a statement about your list and your timing, not your agent — the agent has not spoken yet.
Low and falling means the list is aging, the numbers are wrong, or your caller ID is being filtered. Low on a fresh list means your timing is wrong. It is the first number to look at because everything downstream multiplies by it.
Average call duration
The most under-used number on the page, and the one that costs money directly.
Rising duration with flat outcomes means the agent is talking too much — a prompt fix worth real money at volume. Very short average duration means people are hanging up early, which is a greeting problem. Either way it points at a specific edit.
Rating distribution
Not the average — the shape. A healthy distribution has a spread. If everything clusters at 3 and 4, the rubric is not discriminating and your shortlist is arbitrary, which is worse than having no shortlist because it looks trustworthy.
The ones that mislead
Total calls. Feels like progress, measures activity. Four hundred calls that produced twelve interviews is worse than a hundred that produced twenty.
Average rating. Moves when your list quality changes, not when your process improves. Watch the distribution instead.
Total minutes. A cost, not an outcome. Useful for the wallet, useless for judging whether the programme is working.
The number nobody puts on a dashboard
Cost per useful outcome. Total spend divided by the number of people who reached the next stage — interviews booked, demos scheduled, payments made.
It is the only figure that answers "is this worth doing", and it usually has to be worked out by hand because it needs data from your side as well as ours. Do it once a month. It is the number to take into a budget conversation.
A weekly routine
Connect rate — is the list still good? Average duration — is the agent getting chattier? Rating spread — is the rubric still discriminating? Three numbers, five minutes, and each one points at a specific thing to change.