Call center availability testing: four levels from SIP ping to IVR verification
You give PingVo a phone number and how often to test it. From then on we call it the way your customers do, and we tell you when that stops working. Nothing is installed on your PBX and nothing changes at your carrier. This is contact center monitoring that covers any number answering automatically — IVR menus, auto-attendants, conference bridge dial-ins, hotlines and voicemail-fronted lines: if it answers by itself, we can test it.
L0 — Line check (SIP OPTIONS monitoring) — is the trunk alive?
The cheapest question to ask is whether your provider is answering at all, so we ask it constantly. Every minute or two PingVo sends a tiny status request to the provider handling your numbers and waits for a reply. It costs nothing, places no call, and rings no phone — so we can run it far more often than a real call.
This check is deliberately per-provider rather than per-number: one probe covers every number that provider carries. It is an early warning and a diagnostic, not a verdict — a failed line check on its own never means your number is down. What it does is tell us to start testing harder, and it tells you whether a failure that follows is your equipment or your carrier's.
Technically this is SIP OPTIONS monitoring: a dialog-less OPTIONS request sent to each route's gateway, so a single probe covers every number riding that SIP trunk. Any well-formed SIP response proves the SBC is reachable and processing signalling, which makes it a genuine SBC health check that costs nothing at the carrier.
Want to see one? Run our free SIP OPTIONS check against your own SBC: it sends a single OPTIONS request and shows the response code and round-trip time, and look up any code in our SIP response code list.
L1 — Real call (SIP INVITE test calls) — does it connect?
On your chosen schedule, PingVo dials the actual number over a real carrier and waits for it to be answered. This is the check that catches the failures that matter most and hide best: a number that was ported and never re-provisioned, a trunk that rejects calls at capacity, a routing change that sends callers to a dead endpoint, an expired registration.
Because it is a real call over a real carrier, it exercises exactly what your customer exercises. If it connects for us, it connects for them.
Under the hood this is a SIP INVITE test call placed over a live carrier to the actual DID, not a synthetic probe against a lab endpoint. That is what makes it DID monitoring in the strict sense: it is the only check that proves number routing, PBX answer and trunk capacity together, because a trunk that happily answers OPTIONS can still reject the call that follows.
L2 — Audio (RTP media verification) — is there sound, or dead air?
A call that is answered is not necessarily a call that works. One-way audio and silence-after-answer are the classic invisible faults: every dashboard is green, the call log shows a successful connection, and the customer hears nothing and hangs up.
PingVo listens to what comes back and checks that real audio is actually flowing. If the line answers into silence, that is a failure, and it is reported as one.
Concretely, PingVo verifies the RTP media stream returned after answer and measures its energy, which is what makes one-way audio detection and dead air detection possible at all: the signalling insists the call succeeded, and only the media says whether anyone could actually hear anything.
Every answered test call also gets a synthetic audio quality score from 0 to 100, computed from the decoded audio itself — signal-to-noise, clipping, dropped or repeated audio, and silence gaps — and trended over time on that number's chart, so a line that is slowly getting worse is visible long before it fails outright.
L3 — IVR check (audio fingerprinting) — right menu, right greeting?
The last check is whether the caller hears the correct thing. You record what the number should say once, in a short guided step where PingVo captures the prompt a few times and you approve the best capture. After that, every test compares what plays against what you approved.
This is what catches the embarrassing failures: the holiday message still running in February, an IVR update deployed with the wrong audio, a menu that silently reverted after a maintenance window. It runs on Pro plans and on numbers tested over your own trunks.
The mechanism is audio fingerprinting: PingVo matches an acoustic fingerprint of the greeting that plays against the reference you approved, so automated IVR verification works on the sound itself and survives re-encoding along the carrier path.
This is also why the check is not limited to IVR menus. A conference bridge dial-in, an auto-attendant or a hotline greeting is fingerprinted exactly like an IVR prompt — if the number answers by itself and says something predictable, PingVo can verify it. Conference bridge monitoring and auto attendant testing are the same check, pointed at a different number.
What happens when a check fails
One failed call is not an incident. Networks glitch, and a monitoring tool that pages you for every blip gets muted within a week — at which point it is worse than useless. When a check fails, PingVo tightens the test interval and re-tests to confirm. Only a confirmed, repeated failure opens an incident and sends an alert; when the number recovers, the incident closes and you get the all-clear.
Every incident carries a likely cause, worked out by combining the failed call with what the line checks saw at the same moment: your side, the carrier being down, a carrier routing problem, or an issue with the test itself. Failed calls keep a short recording of what the test actually heard, so you can verify the fault yourself. Alerts go out by email on every plan; Pro adds signed webhooks and a metrics endpoint so incidents reach your own tooling.
Run the tests from where your calls actually originate
By default PingVo tests from our own infrastructure over public carriers — which is exactly right for answering "can the outside world reach this number?".
On Pro you can also run a PingVo agent inside your own network as a container. It tests your internal extensions and your own SIP trunks directly, so you see faults on the segments a public call never touches.
The agent only makes outbound connections to us — nothing needs to be exposed or port-forwarded. It works the same way whether you run it in a branch office, a data centre, or beside your PBX.
Prompt matching runs inside the agent, on the machine that captured the audio. For numbers tested by your own agent, the recording is analysed in your network rather than shipped to us for processing.
Numbers can also be marked so that no audio is ever stored at all — the check still runs and still alerts, it just keeps nothing afterwards.
Frequently asked questions
How do I monitor a SIP trunk?
Point PingVo at the trunk's gateway and it sends a SIP OPTIONS probe every 30 to 60 seconds, which costs nothing because carriers bill answered minutes rather than SIP messages. Pair that with scheduled test calls to a number on the trunk and you see both that the SBC is alive and that calls across it actually complete. Start a free trial and your first trunk is monitored in minutes.
How can I test an IVR automatically?
You record the prompt your callers should hear once, in a short guided step, and approve the best capture. From then on every test call compares the audio that plays against that reference by fingerprint, so a wrong, stale or silent greeting is caught on the next run rather than by a customer. Try it on your own IVR.
How do I detect one-way audio?
One-way audio never shows up in signalling — the call connects, the call log looks clean, and the caller hears nothing. PingVo detects it by verifying the RTP media coming back after answer, so a call that is answered into silence is recorded as a failure and alerted like any other outage. See it on your own numbers.
Can PingVo monitor a conference bridge number?
Yes. Any number that answers by itself can be tested, and a conference bridge dial-in is an ideal case because its greeting is fixed and predictable. The bridge prompt is fingerprinted exactly like an IVR menu, so you find out when the bridge stops answering before a meeting full of people does. Add a bridge number and see it tested.
Does PingVo measure voice quality?
Yes. Every answered test call gets an audio quality score from 0 to 100, derived from signal-to-noise, clipping, dropped or repeated audio and silence-gap analysis, and that score is trended per number so you can see a line getting worse over time. It is a synthetic score computed from the audio we actually heard, not a carrier MOS rating. Start monitoring and the trend builds from the first call.
How often should I test my phone numbers?
Line checks are free to run, so they go every 30 to 60 seconds on every route. Real test calls consume carrier minutes, so most teams run them every 15 minutes on public numbers and every 30 to 60 seconds over their own trunks, where there is no per-minute cost. When a check fails PingVo tightens the interval automatically to confirm before it alerts. Pick a schedule in the trial and adjust it later.
I already have call reporting. Why do I need test calls?
Call reporting tells you about the calls your customers made. It cannot tell you about the calls they could not make. Nights, weekends and holidays are where that gap is widest: no calls is the normal pattern, so a failure hides inside it. PingVo places the calls itself on a fixed cadence, so the answer is the same at 3 a.m. as at 10 a.m. — your line either answered or it did not. Start a free trial and see what your lines do overnight.