VoIP Routes: Origination, Termination, Quality

A VoIP route is the network path a call takes between your switch and the called party’s carrier. Origination routes carry inbound calls from the PSTN to your systems; termination routes deliver your outbound calls to the PSTN. Route quality is largely set by how many networks sit in between — the fewer hops, the better the call.

Every provider sells “routes,” and almost none of them explain what the word means. This guide defines the terms honestly, so you can hear a sales pitch and know exactly what is — and is not — being offered.

SIPNEX is an FCC-licensed carrier that terminates US traffic on its own network — wholesale voice termination is the product this vocabulary describes — so this is the carrier’s-side explanation, bias disclosed.

Origination vs termination: which direction is which

The directions are named from the PSTN’s point of view, which is why people mix them up. Twilio’s Elastic SIP Trunking docs put it cleanly: termination is placing outgoing traffic from your infrastructure to the PSTN; origination is receiving incoming traffic on your numbers, delivered from the PSTN to your infrastructure.

Direction	The call flows	Who buys it
Origination	PSTN → your switch (inbound)	Anyone who needs to receive calls on real numbers
Termination	Your switch → PSTN (outbound)	Dialers, PBXs, resellers sending outbound calls

A memory hook: origination is where the call originates — out on the public network, headed to you. Termination is where your call terminates — someone else’s phone.

Origination is inseparable from numbers: a DID number is the on-ramp, and the origination route is the path inbound calls travel from the PSTN to whatever phone system answers them. Termination is the metered product wholesale buyers actually shop for — per-minute delivery of outbound calls, the core of any wholesale termination relationship. Most operations buy both from the same carrier; they are still two different routes doing two different jobs.

VoIP origination: what the inbound route includes

Carriers call the numbers side of the business origination. The carrier holds DID inventory, and an origination service pairs each number with the inbound route that carries calls to your phone system. When a vendor sells VoIP origination services, you are buying two things at once: the number and the path from the PSTN to your SIP endpoint. Two records ride with that number — its CNAM name and its E911 address — and whether they are included or sold as add-ons is a question to ask before you buy.

Judge an origination service on different evidence than a termination route. Inbound PDD is felt by the caller, not by you, so you cannot time it from your own switch. Check these instead:

• Failover. Where do inbound calls land when your SIP endpoint stops answering — a second IP, a forwarding number, or a fast busy?
• Porting. How long does a port-in take, and who runs the paperwork?
• Inbound caller ID. Does the route pass the calling number through untouched, so your PBX, CRM screen-pop, and call logs see the real number?
• Concurrency. Are inbound channels capped per number, per trunk, or not at all?

What makes a VoIP route good

Four properties separate a good route from a cheap one, and every one of them degrades as hop count rises.

Hop count. A direct route hands your call from your carrier’s switch to the terminating network. A brokered route passes it through resellers first — each running its own routing logic, each a fresh place to fail. Everything below follows from this number.

Setup delay on a route. Post-dial delay (PDD) is the gap between dialing and hearing ringback. Bandwidth’s glossary notes it is more common on wholesale termination (LCR) products. The reason is simple: multiple carriers each take a few seconds to acknowledge they can complete the call. Calls under about 7 seconds of PDD are generally considered acceptable. Dialer operators hold routes to far tighter standards than that: long, variable PDD wrecks predictive pacing.

Audio quality. As an engineering matter, every added network is another chance for the media path to be transcoded, congested, or re-timed. That reasoning is general rather than a published carrier finding, but the pattern operators see is consistent: jitter, latency, and MOS problems multiply on paths nobody fully controls.

CLI integrity. Caller-ID trust erodes across long carrier chains. It is enough of an industry problem that i3Forum, the international carriers’ body, runs a “Restore Trust” initiative and published a 2025 report on restoring CLI origination authenticity. Routes that manipulate or lose CLI are how legitimate campaigns end up labeled as spam.

Attestation survival. Under STIR/SHAKEN, full A-level attestation comes from a provider that knows its customer and their right to the number. SIPNEX signs outbound calls at A-level with its own SP certificate; analytics engines treat partial (B) attestation as a lower-trust signal, though no carrier publishes a specific answer-rate gap.

The route types buyers hear about

Direct (carrier-direct) routes. Your provider is a licensed carrier delivering traffic on its own network and interconnects, not reselling someone else’s. Fewest hops, most accountability — when something breaks, the party you pay is the party who can fix it.

Wholesale A–Z routes. In international wholesale, providers sell rate decks covering destinations “from A to Z,” each destination priced per minute. It is a legitimate market with honest and dishonest corners alike. SIPNEX’s core termination network is the United States, and it does not publish an A–Z international deck — if a vendor quotes you one, you are shopping a different market than this page describes.

Least-cost routes. LCR sends each call to the cheapest supplier that can complete it, re-ranked as rate decks reload. Applied across vetted routes it is rational purchasing; applied blindly it is how cheap minutes become unanswered calls. Our least cost routing guide covers the machinery in depth.

Grey routes. Termination paths that dodge the destination network’s licensing or interconnect arrangements. They surface in the market as suspiciously cheap decks with mangled caller ID and routes that die without notice. Know the term so you can recognize the symptom — no serious operation builds on them.

CLI routes vs non-CLI routes

Wholesale rate decks often split termination into CLI and non-CLI routes. A CLI route carries the calling number end to end, so the person you dial sees the number you sent. A non-CLI route (sometimes written NCLI) strips or replaces the caller ID. The far end may see a substituted number, a generic one, or no caller ID at all. TelcoBridges’ wholesale-termination primer is blunt about the trade: a non-CLI route “is cheaper but hurts answer rates and, in many markets, runs into regulatory and anti-spoofing scrutiny.”

For US outbound traffic the choice is not really a choice. The signed STIR/SHAKEN token names the calling number, so a route that swaps that number in transit leaves a signature that no longer matches what rings. Your caller ID reputation rides on that same number. The highest-quality routes are CLI routes: the number you send is the number that displays, on every call, with the attestation intact.

How VoIP routes are priced

Termination is priced per minute against the destination, and the billing increment quietly decides what a “rate” costs in practice. Twilio and Telnyx bill voice at 60/60 — every call rounds up to a full minute. SIPNEX bills dialer traffic at 12/6: a 12-second minimum, then 6-second increments, so a 45-second call bills as 48 seconds rather than 60.

SIPNEX publishes its rates, and the band it publishes is a ceiling rather than a floor: clean traffic profiles are quoted against actual CDRs, below that published band. That is the honest way to buy a route. Send your real traffic picture and get a quote against it. Skip any vendor whose pricing only exists in a spreadsheet they will not stand behind.

Evaluating a route without a rate table

Rates are the last thing to compare, because a route’s worth is measurable before you commit volume.

Run test calls. Dial numbers you control across the destinations you care about — mobile and landline, several states. Time the PDD yourself, listen to the ringback, confirm the audio, and check what caller ID the far end actually displays.

Watch ASR and ACD on a pilot. Send a small slice of live traffic and read the completion stats in real time. ASR and ACD are the route’s report card: a sagging answer-seizure ratio or an erratic average call duration on identical lead lists means the route, not the list, changed.

Ask how the calls are routed. Do you terminate on your own network or broker across suppliers? How many carriers sit between your switch and the called party? Does the routing rotate when rate decks reload? A provider proud of its routes answers in one sentence. A provider reselling an LCR engine changes the subject to price.

Frequently asked questions

What is a VoIP route?

A VoIP route is the network path a call follows between a provider’s switch and the called party’s carrier. Origination routes bring inbound calls from the PSTN to your systems; termination routes carry your outbound calls to the PSTN. “Route” describes the path and every network in it — which is why two providers selling the same destination can deliver completely different call quality.

Are origination and termination different routes?

Yes — they are separate products moving calls in opposite directions. Origination delivers inbound calls from the PSTN to your infrastructure via your phone numbers; termination delivers your outbound calls to the PSTN. A carrier can be strong at one and weak at the other, so evaluate each direction on its own evidence even when buying both from one provider.

Why do multi-hop routes add post-dial delay?

Because every carrier in the chain takes time to accept the call before passing it on. Bandwidth’s glossary notes PDD is more common on wholesale termination (LCR) products, where each of several carriers “can take a few seconds to acknowledge their ability to complete the call,” with under 7 seconds generally considered acceptable. Direct routes skip those intermediate acknowledgments, which is why they ring faster and pace predictive dialers more predictably.

How do I test a VoIP route before committing traffic?

Three steps. First, place test calls to numbers you control and check PDD, audio, and the caller ID actually displayed. Second, run a small live pilot and watch ASR and ACD against a known lead list. Third, ask the provider directly whether it terminates on its own network or brokers across suppliers. A route that survives all three checks is worth a rate conversation — one that fails any of them is not, at any price.

What is the difference between CLI and non-CLI routes?

A CLI route delivers the caller ID you send; a non-CLI route does not guarantee it. On a CLI route the calling number travels end to end and displays on the called phone. On a non-CLI (NCLI) route the number may be replaced with a generic or substituted one, or dropped entirely, in exchange for a lower per-minute rate. For signed US traffic, use CLI routes — a swapped number undermines the STIR/SHAKEN signature and the reputation of the number you dialed from.

How do I evaluate an origination service?

Test the inbound path, not just the number. Dial your new DIDs from mobile and landline phones in several states and confirm each call reaches your phone system with the caller’s real number intact. Then check failover by taking your SIP endpoint offline and watching where calls land. Finally, ask how port-ins are handled and whether inbound channels are capped. A service that passes those checks is ready for live traffic.