Post-dial delay is the gap between the moment a call is placed and the moment the caller hears ringing. Nothing appears to happen during it. That silence is where a surprising amount of money goes.
On a predictive dialer the cost is obvious: every second of PDD is a second an agent sits waiting. On manual dialing it is subtler. People assume the call failed and hang up. Your carrier's answer statistics look fine, because from the network's point of view the call was never answered — and technically that is true.
What is actually happening in that gap
When your switch sends an INVITE, it starts a chain. Your carrier looks up
where the number should go, picks a route, and forwards the request. That carrier may do the
same thing again. And again. Each hop adds signalling time, and the delay is cumulative.
The gap ends when something comes back — usually 180 Ringing, sometimes
183 Session Progress carrying early media. That is the moment your caller finally
hears something.
Four things drive the number up:
- Hop count. The single biggest factor. A direct interconnect to the terminating network might involve one hop. A cheap route resold four times involves four, each adding its own lookup and forwarding time.
- Least-cost routing that tries and fails. If a carrier's first choice rejects the call, their switch tries the next, then the next. Each failed attempt is time your caller spends listening to nothing. This is very common on cheap routes and almost invisible unless you look at signalling.
- Number lookups. Ported-number databases and ENUM queries take time, particularly for mobile numbers in markets with heavy porting.
- Distance and peering quality. Physics sets a floor. A call from a server in Frankfurt to a mobile in Lagos cannot be instant. Poor peering makes it considerably worse than the floor.
What good looks like
PDD varies legitimately by destination, so a single target number is misleading. As a rough guide for a route that is behaving:
| PDD | What it usually means |
|---|---|
| Under 1s | Direct or near-direct route. What you are paying for on a premium CLI route. |
| 1–2s | Normal for most international destinations. No cause for concern. |
| 2–4s | Several hops, or a distant destination. Tolerable on manual dialing, painful on a dialer. |
| 4–8s | Something is wrong. Usually failed route attempts before one succeeds. |
| Over 8s | Most callers have already hung up. Whatever you are saving per minute, you are losing on connects. |
Compare like with like. UK landline and Nigerian mobile will never produce the same number, and a carrier who quotes you a single "average PDD" across all destinations is quoting you a number that means nothing.
Working out whose problem it is
Before you blame the carrier, rule out your own side. Three checks, in order:
- Is it one destination or all of them? If PDD is high everywhere, the problem is likely your switch, your internet connection, or your codec negotiation. If it is high on one prefix, it is the route.
- Did it change, or has it always been this way? A sudden jump on a route that was fine last week almost always means your carrier changed their upstream. They may not have told you. They may not know.
- Does it vary by time of day? PDD that climbs during peak hours points at capacity — either congestion on the route, or a carrier whose primary path fills up and starts overflowing to slower alternatives.
Reading it from the signalling
If you have access to SIP traces on your switch, the measurement is exact: the time between
your INVITE and the first 18x response. Most softswitches log this,
though not all show it prominently. In Asterisk, sip set debug or a
pcap through sngrep will show you the timestamps directly.
What you are looking for is not just the total. Look at whether anything came back at all
before the ringing — a 100 Trying arriving quickly followed by a long silence
tells you your immediate carrier accepted the call and then waited on someone else. That is a
routing problem upstream of them, and it is worth telling them so.
The trade-off nobody states plainly
Cheap routes are cheap partly because they are resold several times. Each resale adds a hop, and each hop adds delay. This is not a defect in the route; it is what the route is.
So the real question is not "how do I get cheap minutes with low PDD" — it is which one your particular traffic can afford to lose. A campaign with expensive agents sitting idle cannot afford delay. A low-volume support line probably can.
Work out what a second of PDD costs you in agent time, then compare it to the price difference per minute. Often the expensive route is the cheaper one.
If you are testing a new carrier, measure PDD during the trial, not after. Ask for the route type in writing, dial your real destinations at your real volume, and watch what happens at peak hour rather than at nine in the morning when nothing is loaded.
Want this checked on your own routes?
Send us the destinations you dial and we will look at what you are actually getting, whether or not you end up buying from us.
Ask us
