Your connect rate is probably a lie

Here is a test. Open whatever you use to track calls and find your connect rate. If it is above about 60% on cold outbound, it is almost certainly measuring the wrong thing.

Where the error comes from

There are two clocks in telephony and they are not the same.

The bridged clock starts when somebody answers. This is what a softphone reports, and duration above zero genuinely means a human picked up.

The dialer clock starts when dialling starts. Ringing counts. Voicemail counts. A number that rings out for eighteen seconds and goes nowhere produces an eighteen second call.

Most CRMs define connected as duration > 0, which is correct for the first clock and meaningless for the second. If any of your call data came from a dialer export, and almost everybody's did at some point, you are counting ring outs as conversations.

On our data, 1,263 of 2,694 outbound calls were between one and nineteen seconds. Under the naive definition every one of those was a connection.

The fix

Use a duration floor of twenty seconds.

It is long enough to exclude every ring out and voicemail drop, and short enough to include a genuine "not interested, goodbye", which is a real connection even though it is a fast one.

On the same data that produced a 95% connect rate, twenty seconds produces 50%.

Then add a second threshold at sixty seconds and call it a conversation. Now you have two numbers that mean different things: did somebody pick up, and did somebody stay. The gap between them is the single most diagnostic figure on an acquisitions desk, because it isolates your opener from your list.

• Low connects, high conversation rate among those connects: your list or your data is bad.
• High connects, low conversation rate: your first fifteen seconds are bad.

One of those is fixed by buying better data. The other is fixed by listening to five recordings this afternoon. They look identical on a dashboard that reports one number.

Why this matters more than it sounds

A metric that is obviously wrong gets ignored, which is survivable. A metric that is quietly wrong gets trusted, and every decision downstream of it inherits the error. If your connect rate says 95%, you will conclude your list is excellent and your problem is closing, and you will spend three months coaching the wrong thing.

Check the definition before you trust the number. It takes ten minutes and it is the highest return ten minutes available to most desks.