Sales call latency: why prospects say hello twice

Sales call latency is the delay between a person speaking and the other side hearing them. Keep it under 150 ms one way and most people will not notice it. As it climbs toward 400 ms, the ITU’s planning limit, reps and prospects start talking over each other, and the call feels awkward before anyone has said anything wrong.

A rep in Bengaluru calls a finance director in Sydney. The director picks up and says “Hello?” Nothing comes back for what feels like a beat too long, so they say it again, a little sharper. By the time the rep’s opener arrives, it lands on top of the second hello. Neither person did anything wrong. The audio was simply late.

What the 150 ms and 400 ms numbers mean

The reference most carriers and engineers use is the ITU’s Recommendation G.114, which covers one-way transmission time. It measures delay “mouth to ear”: from the moment someone speaks to the moment the other person hears it.

G.114 makes two points that matter for a sales floor:

One-way delay	What G.114 says	What it feels like on a call
Under 150 ms	Most applications will experience “essentially transparent interactivity”	A normal conversation
150–400 ms	Acceptable for planning; the effect should be estimated with the E-model (G.107)	Noticeable on fast back-and-forth, such as an opener and a quick objection
Over 400 ms	Recommended not to exceed for general network planning	People talk over each other and repeat themselves

The recommendation also says users expect local and regional calls to be “completely delay-transparent”. A prospect may forgive a slight lag on a call they know is international. They will not forgive it on what looks like a local call, which is exactly what local presence dialing makes your calls look like.

Where the delay comes from

Latency is a sum. Each part of the path adds a little, and the parts you control are usually bigger than distance.

1. Distance. G.114 lists optical fiber at about 5 µs per kilometer. 10,000 km of fiber adds about 50 ms. That is real, but it is rarely the whole story.
2. The route. Audio does not travel in a straight line. If your calling platform relays media through a server in another region, the audio travels there and back on every word.
3. Packetization and codecs. Voice over IP cuts speech into packets, often 20 ms each, and encodes it. G.114’s VoIP appendix uses reference terminals that add 50–80 ms in total.
4. The jitter buffer. Packets arrive unevenly, so the receiving end holds them briefly to play them in order. For planning, G.114 says to assume a jitter buffer adds half its peak size. A buffer built for 50 ms of variation adds about 25 ms.
5. The last few meters. A rep on crowded Wi-Fi, a wireless headset, or an overloaded laptop can add delay and packet loss that no carrier can remove.

For context, G.114’s VoIP appendix estimates that well-engineered routes of 5,000 km or less are likely to stay under 150 ms. Routes of about 10,000 km come in near 225 ms, and the ITU’s worst-case 27,500 km route comes in at just over 300 ms.

A worked example: the detour costs more than the distance

Here is an example with round numbers, to show why the route matters more than the map. It uses G.114’s 5 µs per kilometer for fiber and its reference terminal delay. The distances are illustrative, not measured.

	Direct route	Relayed through a far-away media server
Fiber path	9,000 km	9,000 km + 2 × 12,000 km detour = 33,000 km
Distance delay (5 µs/km)	45 ms	165 ms
Terminals (codec, packets)	80 ms	80 ms
Jitter buffer (half of 50 ms)	25 ms	25 ms
One-way total	about 150 ms	about 270 ms

Same rep, same prospect, same headset. The relayed call adds 120 ms in each direction, so the gap between a question and its answer grows by about a quarter of a second. On a cold call, that is the difference between a crisp opener and a “sorry, go ahead”.

Latency is not the same as connect time

Two different delays get blamed for the same “hello?”:

• Audio latency is the delay on every word, for the whole call.
• Connect time is how long it takes, after the person answers, for a rep to be on the line. Dialers spend this time deciding whether a person, a voicemail, or a phone menu answered.

US rules care about the second one. The FTC’s Telemarketing Sales Rule treats a call as abandoned when the telemarketer does not connect a person to a sales representative within 2 seconds of their completed greeting (16 CFR 310.4(b)(1)(iv)). The safe harbor allows no more than 3% of calls answered by a person to be abandoned per campaign, measured per 30-day period. For how that rate is counted, see how to calculate your abandoned call rate.

Parallel dialing raises the stakes on both delays, because the rep is waiting while several lines ring. If you are weighing it, our guide to what a parallel dialer is covers how the first answer reaches the rep.

A 15-minute check you can run today

You do not need a network engineer to find out if latency is costing you conversations.

1. Listen to 10 recordings from yesterday’s answered calls. Count how many start with the prospect saying hello twice, or with both people talking at once.
2. Split the delay. On each one, note the silence between the prospect’s first hello and the rep’s first word. Over 2 s is a connect-time problem. A short gap that still feels laggy is audio latency.
3. Call your own mobile from the dialer, from the same desk, and say a few words. Then call a colleague in the country you sell to. Talk over each other on purpose and listen for the lag.
4. Move the rep to a wired connection for an hour and compare. If the problem disappears, it was the last few meters, not the route.
5. Ask your provider two questions: where does call audio for my reps and my prospects’ countries actually flow, and what one-way delay do you target on those routes?

If the answer to the last question is vague, that tells you something too.

How DialSavvy handles this

DialSavvy is the carrier for your calls, so the route your audio takes is ours to plan, not something you configure. On our countries page, meeting a latency target is one of the gates before parallel dialing turns on in a country; until then, that country stays on power or single-line dialing. More on how numbers and routes work is on the telephony page.

If prospects keep saying hello twice, book a demo and bring a few of the countries you call. We’ll place test calls on those routes with you and listen for the lag together.