For a decade now, IP telephony has been sold as an obvious saving. In practice it's a decision worth thinking through — because a botched migration can cost more than the old Panasonic PBX it was meant to replace.
We mostly see two extreme scenarios. In one, somebody shows the owner a deck that says "you'll save 70% on your phone bill", the company buys 3CX, plugs into the first SIP trunk it finds on the internet, and spends six months fighting crackling on the line. In the other, somebody once said VoIP "fails when the power goes out", and the company spends another five years paying 2,000 PLN a month for a PBX that serves 18 people and can't forward a call to a mobile.
The truth, as usual, sits in the middle. IP telephony is good technology, but it isn't magic. You need to know what you're buying — and when it's worth buying at all.
When the old PBX is fine
If you have a 5-person office, two outside lines and nobody works remotely — leave well enough alone. A PBX that's worked for twelve years needs no investment and no training. Until the operator pulls your ISDN lines (and they're pulling them, methodically), there's nothing to chase.
Migrating to VoIP starts to make sense when one of four signals shows up:
- Your phone bill is more than 1,500 PLN per month net.
- Staff take work calls on personal mobiles because "nobody is in the office anyway".
- Sales reps ask for call recording (disputes, training) and the PBX can't do it.
- The operator has notified you that your copper lines are being decommissioned next year.
Then it's worth running the numbers. Before that — not really.
What you actually buy in a migration
VoIP isn't just "cheaper calls". Cheaper calls are a bonus. The real value sits in three things the old PBX can't do.
First, CRM and ERP integration. Click-to-dial from a customer record. Pop-ups with caller information. Call notes saved directly into the contact history. For sales reps, that's the difference between "I remember talking to them somewhere" and "Friday at 14:32 we agreed on a 5% discount for order XYZ". In Comarch Optima, enova365 or Subiekt GT this can be wired up — sometimes natively, sometimes through 3CX and a CRM connector in the middle.
Second, work from anywhere. A rep on the road takes calls on the company number through a mobile app. The customer never sees a private number. Back at the office, the calls don't get lost inside a personal app.
Third, recordings and statistics. Who spent how long on calls, how many were abandoned, what the average response time looks like. Basic numbers the old PBX doesn't surface, that genuinely help you run a customer-facing team.
The bill savings are a bonus. The math: what you pay today for line rentals and minutes, against what you'll pay for the SIP trunk, 3CX (or another softswitch) licences and the server/cloud. Monthly bills typically drop 50-70%, but only the specific numbers tell you how much.
A concrete example: a 25-person design firm
At one of our clients — a 25-person architectural design firm near Warsaw — the picture looked like this. An old Panasonic KX-TDA PBX, 8 ISDN lines, operator costs of 1,850 PLN per month net. The PBX worked, but: no recording, no mobile app, no integration with their CRM (they use Pipedrive). The designers were increasingly working in the field — and giving clients private numbers because "there's no other way".
Our analysis came out in favour of migrating to a self-hosted 3CX (a small server in the office), a SIP trunk from a domestic operator and 28 user licences. One-off investment: licences, VoIP phones (we kept most of the DECT handsets, replaced some desk phones), implementation work — 42,000 PLN net in total.
After the migration, the operator bill dropped to 510 PLN per month (SIP trunk, numbers, minutes). Add annual 3CX support and the server amortised monthly — about 220 PLN. Total new running cost: 730 PLN per month instead of 1,850. Savings: 1,120 PLN per month = 13,440 PLN per year. ROI on bill savings alone: 3 years, 2 months.
That isn't a spectacular return. The spectacular part is the side effect: the Pipedrive integration meant reps stopped guessing who was calling. Average time to handle a phone-driven quote enquiry dropped from 14 to 6 minutes. Recordings settled two disputes that would otherwise have been "his word against ours". The client didn't bother converting that into złoty, but said "had I known, I would have done this three years earlier".
Three things demos quietly skip
VoIP sellers don't lie, but they go quiet on a few things that show up painfully later.
Internet quality. VoIP lives on the internet. If you have one circuit, no backup, and the fibre in your street is cut once a quarter — your calls go down with it. In practice, for 15+ users you want two circuits (fibre + LTE failover, for instance) and a router that can switch between them. That's 200-400 PLN a month more, but without it your first outage breaks the whole business case.
LAN QoS. One misconfigured switch, someone running a backup down the same cable mid-afternoon — and the calls start crackling. The office LAN, where "the internet works" used to be enough, now needs structure: VLANs, QoS, decent switches.
Backup power. The old analogue PBX worked through a power outage because the lines were powered from the operator's exchange. VoIP isn't. You need a UPS for the router, the switches and the phone server. Otherwise your first hour-long outage = an hour without phones. Cost: 1,500-3,000 PLN one-off, and nobody mentions it on the demo.
How to do it without pain
A sequence that works:
- Add up today's cost. Operator invoices for the last twelve months. How much in subscriptions, how much in minutes, how much in extras (forwards, hotlines). Without that, there's no comparison.
- Inventory who actually makes calls. Often it turns out that of 30 phones on desks, only 12 are really used. The rest are relics.
- Pick a softswitch. 3CX is popular in Poland, has sensible licensing and plenty of local integrators. Alternatives: FreePBX (cheaper but more skill required), operator-hosted cloud telephony (less flexible, but zero infrastructure on your side).
- Test the network before migrating. Latency, jitter, packet loss. If the numbers are bad — fix the network first, then move the phones. Not the other way around.
- Port numbers deliberately. Porting a landline to another operator takes 14-30 days. Plan it, because "we'll switch on Friday evening" doesn't work here.
What not to buy
A few things that sound great but rarely pay back.
Bells and whistles like "AI bot for the call centre" — a 25-person firm doesn't need it. Premium-brand desk phones — mid-range Yealink or Fanvil units do the same job. The cheapest unbranded Chinese handsets — 80 PLN saved per unit, years of support pain. Per-minute cloud subscriptions sold as "cheaper" — at 25 users, a SIP trunk with a minute bundle is almost always cheaper.
What next
VoIP in an SME isn't a "wow" decision. It's an "I ran the numbers and it makes sense" decision. If your operator bill is over 1,500 PLN, people work remotely, and you feel the old PBX is holding you back — do the math. If everything works and the operator hasn't announced the end of your lines — you can wait another year.
If you don't know which side you're on, half an hour with someone who does these migrations for a living will save you a lot of guessing.
VoIP isn't a cheaper version of the old PBX. It's a different technology — with different strengths and different traps.