ASR and ACD Explained: How to Judge a VoIP Route

Two numbers come up in every conversation about VoIP route quality: ASR and ACD. Carriers quote them, buyers demand them, and both are easy to misread. This guide explains how they're calculated, what healthy values look like for different traffic, and the patterns that suggest a route is being gamed.

ASR: Answer-Seizure Ratio

ASR is the percentage of call attempts that are answered.

ASR = answered calls ÷ total call attempts × 100

An "answered" call is one that received a 200 OK. A "seizure" is any attempt the carrier accepted and tried to route. If you sent 1,000 calls and 450 were answered, your ASR is 45%.

What ASR measures — and what it doesn't

ASR mixes two very different things:

• Network failures — the route couldn't reach the destination, timed out or returned an error.
• Human behaviour — the person was busy, didn't pick up, or rejected the call.

A route can be technically perfect and still show a low ASR if you're calling people who don't answer. That's why ASR is only meaningful when you compare routes on the same traffic, at the same time.

NER: the cleaner number

Many carriers also track NER (Network Effectiveness Ratio). It counts busy, no-answer and rejected calls as successes, because the network did its job — it reached the phone.

NER = (answered + busy + no answer + rejected) ÷ total attempts × 100

NER on a good route is usually well above 95%. If ASR is low but NER is high, the problem is your audience or your timing, not the route.

ACD: Average Call Duration

ACD is the average length of answered calls, usually in minutes.

ACD = total billed minutes ÷ answered calls

If 450 answered calls added up to 1,350 minutes, the ACD is 3.0 minutes.

ACD tells you whether calls are real conversations. When a route degrades — poor audio, dropped calls, one-way audio — people hang up sooner and ACD falls. It's often the earliest warning sign you'll get.

Healthy ranges by traffic type

There's no universal "good" ASR or ACD. It depends heavily on what you're sending. These are rough ranges we see on clean CLI routes:

Mobile-heavy destinations generally show lower ASR than landlines. Time of day matters too: business traffic answers far better in local working hours.

Reading the numbers together

ASR and ACD are most useful side by side:

Patterns that signal a route is being gamed

Some vendors manipulate routes to look better on paper or to bill more minutes. These are the patterns to watch for.

False Answer Supervision (FAS)

The route returns 200 OK before the callee actually answers — sometimes during ringing, sometimes playing a fake ringback or a recorded message. You're billed from the false answer.

Signs: ASR that's suspiciously high for the destination, many very short calls, billed duration that's longer than your own measured talk time, and customers saying "the line was ringing but I was charged".

Late or missing BYE

The far end doesn't release the call promptly when the callee hangs up, so billing keeps running.

Signs: ACD drifts upward, a cluster of calls end at round numbers (such as exactly 60 or 120 minutes), or call durations don't match your PBX's own records.

Route switching to SIM boxes

A "CLI" route is quietly moved onto a SIM-box gateway to save money. Callers see the wrong number and hang up.

Signs: ACD drops sharply over a few days, more complaints about wrong caller ID, rising post-dial delay. Our CLI vs non-CLI guide explains why this happens.

Short-call stuffing

Unwanted or artificial short calls are mixed into traffic to make a weak route look busy, or to hit volume commitments.

Signs: spikes of calls lasting a few seconds that don't match your own campaign activity.

How to measure it yourself

Don't rely only on your carrier's reports. Your own CDRs are the source of truth.

1. Export CDRs from your PBX or softswitch with: start time, answer time, end time, destination, disposition and SIP response code.
2. Group by destination prefix and route. Mixing countries hides problems.
3. Calculate ASR, NER and ACD daily, and keep at least four weeks of history to spot trends.
4. Compare billed duration with your own measured duration. Differences of more than a second or two per call deserve a question.
5. Look at the distribution, not just the average. A histogram of call lengths will show FAS (a spike at a few seconds) or stuck calls (a long tail) instantly.
6. Track SIP response codes. A rise in 503 or 408 points to network issues; a rise in 486 or 480 points to the audience. See SIP error codes.

If you use MagnusBilling, VOS3000 or another softswitch, most of these figures are available per route in the reports section.

What to ask a carrier

When evaluating a new wholesale VoIP route, ask:

• What ASR and ACD do you see on this destination, on traffic like mine?
• Is this route direct, or through other carriers? How many hops?
• Is caller ID guaranteed end to end?
• How do you detect and handle FAS?
• What's your billing increment, and do you bill from 200 OK?
• Can I have test credit to send real traffic for a few days?

A carrier that answers these clearly, and offers a real test, is usually one you can work with.

How we use these numbers

Our NOC monitors ASR, NER and ACD on every route, by destination, around the clock. When a route drops outside its normal band we investigate and, if needed, move traffic to a backup before customers notice. We publish typical figures for our main destinations on the network page, and you can request per-destination stats for your own traffic from support.

Summary

• ASR tells you how often calls are answered — but it's heavily affected by your audience.
• NER isolates the network's part, so use it to judge the route itself.
• ACD tells you whether calls are real conversations, and is the earliest sign of quality trouble.
• Always compare routes on the same traffic, and check your own CDRs against your carrier's.
• Unusually good numbers deserve as much scrutiny as bad ones.