Dead air is the silence between the moment a prospect says "hello" and the moment they hear a human voice. On a cold call it is the single fastest way to lose a pickup you worked hard to get. In our benchmark of 1,000,000 B2B cold calls in France, the average connect rate is 11.3%. Every one of those connects is rare and expensive, and a two-second silence hands it straight back to the prospect, who assumes a robot is calling and hangs up.

This article explains where connection delay comes from, which part of it the prospect actually notices, how to measure it on your own calls, and what to fix first.

Connection delay is two different problems

People use "connection delay" for everything between pressing dial and talking. In practice there are two delays, and only one of them costs you meetings.

Before the answer. Between the dial and the first ring there is carrier routing, number lookup, sometimes an international hop. That can take two to five seconds. The prospect never hears it. It costs your reps patience and a few seconds per attempt, nothing more.

After the answer. Between the prospect's "hello" and the rep's first word. This is the delay the prospect experiences. It is the only one that changes the outcome of the call, and it should be under a second.

When a sales team complains that "the dialer is slow", the first job is to find out which of the two they mean. The fixes have nothing in common.

Why dead air after hello is so expensive

Think about what a prospect does when they pick up an unknown number. They say hello, they wait, and they have already decided within a second or two whether this is a person or a machine. Years of robocalls have trained everyone, in France as much as in the US, to hang up on silence.

The cost compounds. A rep who makes 120 dials in a session on a 11% connect rate gets about 13 pickups. If dead air loses three of them, the rep lost a quarter of their conversations before saying a word. No opener, no script and no coaching can recover a call that ended before it started.

There is a slower cost too. Prospects who picked up once to silence are less likely to pick up that number again. Dead air trains your own market not to answer you.

What eats the second after hello

Three things usually consume the budget.

• Answering machine detection. The dialer listens to the first syllables to decide whether a human or a voicemail answered. A slow detector adds half a second to a second before it routes the call to a rep. A detector that is both fast and accurate is the main engineering difference between dialers.
• Rep availability. In a parallel session, if two prospects answer at the same time and only one rep is free, the second one hears silence. Dialing too many lines for the number of reps is the most common structural cause of dead air.
• Audio setup. The time to bridge the rep's audio into the answered call. Browser-based softphones on a bad Wi-Fi connection add latency here that no dialer can hide.

In our experience the first two are design choices and the third is a laptop and network question. All three are measurable.

How to measure dead air on your own calls

Do not trust what reps feel. Measure it on a controlled test.

1. Put three colleagues' mobile numbers in a test list. Use real mobiles on real carriers, not desk phones.
2. Run a session exactly as reps do: same dialing mode, same number of lines, same browser.
3. Each colleague answers with a normal "hello" and starts a stopwatch at the end of the word. They stop it when they hear the rep's first word.
4. Repeat ten times per mode. Change one variable at a time: lines, headset, Wi-Fi versus Ethernet.

Record the median, not the average. One bad outlier should not drive the decision. A median under one second is good. Between one and two seconds you are losing pickups. Above two seconds the dialer or the setup needs to change before you add volume.

Match the symptom to the owner

Once you know which interval is slow, the fix has a clear owner.

• Dial to first ring is slow: carrier or routing. Check whether international calls go through a local point of presence. Ask the vendor, it is their job.
• Hello to rep voice is slow and consistent: answering machine detection or audio bridging. Vendor question.
• Hello to rep voice is slow only at peak: too many lines per available rep. Sales ops question. Drop one line and measure again.
• Slow only for one rep: their network or headset. IT question, two minutes to fix.

Name the interval before escalating. "The dialer is slow" gets you a shrug. "Median 1.8 seconds from hello to voice on three lines, 0.7 on two lines" gets you a fix.

The European angle: fewer pickups, less room for error

Connect rates in France sit between 11% and 20% depending on persona and slot, with 4pm at 15.8% and 10am at 14% across personas. That is lower than the pickup rates many US playbooks assume, which means two things for dead air.

First, each pickup is worth more, so a lost one hurts more. Second, with fewer simultaneous answers, a well-built parallel dialer collides less often than the maths suggests, as long as the number of lines matches the number of reps. The teams we see struggling are the ones that copied a "dial eight lines" setting from a US blog with a single rep on the session.

How Sonik approaches connection delay

We built Sonik's dialer around one constraint: the prospect should never hear silence. Answering machine detection runs in the first syllables and hands the call to the rep before the prospect finishes "hello". The number of parallel lines adapts to the number of reps free in the session, so a second answer is held for a free rep instead of dropped into dead air.

Every session shows the rep what is ringing, what connected and what went to voicemail, so a collision is visible rather than guessed. And because the audio runs through the browser, we tell every new customer the same boring thing: Ethernet beats Wi-Fi, and a wired headset beats Bluetooth. Half of the "slow dialer" tickets we have ever received were a coffee-shop Wi-Fi.

If you want to see what sub-second hand-off feels like on your own list, the Sonik dialer page explains the dialing modes.

Conclusion

Connection delay is two problems. The one before the answer is a cost of doing business; the one after the answer decides whether you get to speak. Measure it this week with three colleagues and a stopwatch, find the owner of the slow interval, and do not add a single line of parallel dialing until the median from hello to voice is under one second.