Nova Wholesale BridgeMost comparisons of SIP trunking and PRI stop at the monthly line rental, declare SIP the winner, and move on. That is true as far as it goes, but it is not where the money actually is, and it is not what makes a migration succeed or fail.
Here is the honest version.
What a PRI actually costs you
A PRI is a physical circuit carrying 23 or 30 concurrent channels, depending on where you are. The costs break down roughly like this:
- The circuit itself, a fixed monthly charge whether you use one channel or all twenty-three.
- Channel granularity, you buy in blocks. Need 25 channels? You are paying for 46.
- The hardware to terminate it, a PRI card in your PBX, and a spare, because the lead time on a replacement is not measured in hours.
- Per-minute charges, usually higher than equivalent SIP rates.
- Moves and changes, adding capacity means an engineer visit and a lead time measured in weeks.
The last two are where the real cost hides. The circuit rental is visible on an invoice and easy to compare. The cost of not being able to add ten channels before Monday is not.
Where SIP actually saves money
Capacity matches demand. Channels are provisioned in software. If your call volume triples in December and halves in January, your capacity and your bill can follow it. No engineer visit, no lead time.
Per-minute rates are lower, provided your traffic is running on direct routes rather than being handed between wholesalers. This is the part where “SIP is cheaper” quietly stops being true if you pick badly. A cheap route will beat any PRI on price right up until answer rates collapse and your agents are dialling three times to connect once.
No termination hardware. If your PBX already speaks SIP, there is nothing to buy. If it does not, a session border controller or an ATA is still cheaper than a PRI card, and it is not tied to a single carrier.
Redundancy stops being a second circuit. With a PRI, resilience means paying for a second physical circuit that sits idle. With SIP, calls can re-home to an alternate gateway, or fail over to mobile numbers, as a configuration choice rather than a capital one.
The comparison nobody puts in the table
For a 30-seat office running a single PRI, the monthly saving on line rental alone is real but not dramatic. What changes the arithmetic is everything the PRI made impossible:
- Opening a second location without a second circuit
- Taking calls on a local number in a city you have no office in
- Letting people work from home without forwarding every call to a mobile
- Adding twelve seasonal agents for six weeks
None of those appear as a saving. They appear as things you can now do.
The migration steps most guides skip
This is where projects actually go wrong.
1. Audit your real concurrency, not your seat count. Thirty people does not mean thirty simultaneous calls. Pull ninety days of call records and look at the peak, then add headroom. Most teams over-buy channels because they size against headcount.
2. Check what your PBX actually supports. “SIP capable” covers a wide range. Confirm the codecs, whether it supports TLS and SRTP if you need encryption, and whether its SIP stack has known interoperability quirks. An older Avaya or Cisco system may need a session border controller in front of it.
3. Size your internet connection, and separate the traffic. Voice needs roughly 100 kbps per concurrent call with G.711, less with G.729. The bandwidth is rarely the problem. The problem is a large file upload sharing a link with a customer call. QoS on your own network, or a separate connection for voice.
4. Prove the trunk before you cut anything over. Get it live, route test calls both ways, and confirm caller ID presentation and emergency calling while your old service is still running. A trunk configuration is easy to change; a cutover is not.
5. Plan the emergency services registration. With a PRI, your location was implicit in the circuit. With SIP it is not. Registering the correct address for emergency calls is required by the FCC under RAY BAUM’S Act, and it is the single most commonly forgotten step in a migration.
6. Keep the PRI running in parallel for a fortnight. Overlap costs one extra month of line rental. Discovering a routing problem with no fallback costs considerably more.
When a PRI is still the right answer
Rarely, but it happens. If you are somewhere with genuinely unreliable internet and no viable second connection, a circuit that fails independently of your broadband has real value. And if you are two months from a site closure, migrating is not worth the project time.
Everywhere else, the question is not whether to move but who to move to, and that question is mostly about whose network your calls actually run on.
Working out your own numbers
If you want a comparison against your actual traffic rather than a generic table, send us ninety days of call records and we will size the channels and price the destinations you really call. Talk to our network team, it is a conversation with an engineer, not a quote form.