The 60-second answer
Over 36 months for 22 extensions, unbundled hosted call control usually wins — not on monthly price, but because growth is free inside a band, there is no hardware refresh, and exit cost is near zero. On-premise competes where headcount is flat and IT capacity exists.
The number everyone compares, and why it misleads
Phone systems are sold on a monthly figure, and monthly figures are chosen to compare well. What they exclude is everything that happens between month one and month thirty-six: hiring, hardware ageing, configuration changes, and eventually leaving.
Below is a worked 36-month comparison for a plausible Canadian business — 22 extensions at the start, growing to 28, one location, existing handsets in reasonable condition. Figures are CAD and exclude GST/HST, which applies to all three columns at your provincial rate [4] and is generally recoverable by registrants, so it changes cash-flow timing rather than the ranking.
The three-year comparison
| 36-month cost | Bundled carrier | On-premise owned | Unbundled hosted |
|---|---|---|---|
| Up-front hardware | $0 (carrier-owned) | $9,000 system | $0 (handsets retained) |
| Setup / install | $0 | $3,500 | $750 |
| Platform, 36 months | included in seat rate | $0 | $219 × 36 = $7,884 |
| Lines / trunk | 22→28 seats @ $32 | $140/mo trunk = $5,040 | $140/mo trunk = $5,040 |
| → line subtotal | ~$27,000 | $5,040 | $5,040 |
| Maintenance / patching | $0 | $1,800/yr = $5,400 | included |
| Config change fees | ~$600 | internal time | included |
| Growth to 28 extensions | +$32/seat/mo | licence add-ons ~$1,200 | $399 band from mo 20 |
| 36-month total | ~$27,600 | ~$24,140 | ~$16,314 |
| Exit cost at month 36 | Replace all handsets | Hardware aged out | ~$0 |
Your numbers will differ — trunk pricing, install rates and seat rates all vary regionally. The shape holds, and it holds for a reason worth naming: the two expensive columns are expensive in different ways. Bundled service is expensive because growth is charged forever. On-premise is expensive because you bought an asset and then kept paying to look after it.
The cheapest column is not the one with the lowest monthly rate. It is the one where hiring six people costs nothing.
The tax layer, briefly
Buying a PBX outright is a capital outlay rather than a current expense, so it is not deducted in the year of purchase — the general limitation in section 18 of the Income Tax Act is the statutory root of that distinction [3]. Instead it is written off through capital cost allowance over several years [1], with electronic telephone equipment generally falling in Class 8 at 20 per cent declining balance [2].
The practical effect on the table above is that the $9,000 on-premise purchase is real cash in year one but yields its deduction slowly, while a subscription is ordinarily a current expense deductible in the year incurred. For a business managing cash, that timing difference matters as much as the totals. This is general information rather than advice for your circumstances — your accountant should confirm the treatment.
The five costs quotes leave out
- Growth. Under per-seat pricing every hire adds the full rate permanently. Six hires at $32 is $2,304 a year, forever, and it is the largest single divergence in the table.
- Hardware refresh. An on-premise system has a life. At year five or six it needs replacement, and the replacement is another capital cycle. Three-year comparisons flatter it by ending before that arrives.
- Patching. A PBX in your closet is a server, and federal baseline controls expect supported software, applied updates and controlled administrative access [6]. Someone does that work, and either they are paid or it is not happening.
- Change fees. Ten configuration changes a year at $60 is $1,800 over three years for editing your own greetings. Included admin time is worth more than it looks.
- Exit. The one nobody prices at all.
How the totals diverge over time
The interesting property of the three columns is not where they end but where they cross. In month one the bundled option looks best by a wide margin, because it has no setup cost and no capital outlay — you sign, and phones work. That is a genuine advantage and it is why the bundle sells.
On-premise is worst in month one by roughly $12,500 and stays worst for about eighteen months, at which point its low running cost begins closing the gap. If headcount were frozen it would eventually overtake the bundle somewhere around month thirty. It does not overtake hosted, because the maintenance line never stops and the capital cycle restarts before the advantage compounds.
Hosted starts slightly behind the bundle — a $750 setup against nothing — and passes it in month four, purely on the difference between a flat $219 and twenty-two seats at $32. Every hire after that widens the gap rather than narrowing it, which is the opposite of how per-seat pricing behaves and the reason the final totals are not close.
The lesson for anyone comparing quotes is that a three-year window is the minimum useful horizon. A twelve-month comparison ranks these options in almost the reverse order, which is precisely why twelve-month comparisons are the ones you are shown.
Running the numbers for your own business
Four inputs decide your answer, and you can gather all of them in an afternoon:
- Your true extension count — devices that register, not people employed. It is usually 20 to 40 per cent above headcount once shared handsets and analogue devices are included.
- Your hiring expectation over three years — even a rough figure. This single input moves the bundled column more than anything else.
- A transport-only quote from your carrier — lines with no phone-system features. Without it you cannot separate what you pay for the bundle's two halves.
- The age and condition of your handsets — whether they are yours, and whether they are worth keeping.
Put those four numbers into the table structure above and the ranking usually settles itself in under an hour, without anyone needing to sell you anything.
Exit cost is the row that decides the decade
At month 36 you make another decision, and what it costs to choose differently is determined by what you own.
Handsets you bought outright re-register to any conforming platform, because the signalling standard is open and universal [5]. Handsets supplied inside a bundle typically go back, so leaving means buying twenty-eight desk phones before you have even compared providers. That is not a switching cost, it is a deterrent, and it is doing exactly what it was designed to do.
On-premise sits in between: the hardware is genuinely yours, but by year three it may be old enough that moving it is not worth the effort. You own an asset with declining book value [2] and declining practical value at roughly the same rate.
Where on-premise still wins
Flat headcount, hardware already bought and working, in-house IT capacity, and a site with unreliable internet where analogue fallback matters. In that combination the on-premise column can come out ahead on cash over three years, and the maintenance burden is being absorbed by staff you employ anyway. It is a legitimate answer, and hybrid arrangements — on-premise at the difficult site, hosted elsewhere — are more common than the marketing on either side suggests.
Bottom line
Compare 36 months, including the hires you expect, the hardware cycle you will hit, and the cost of leaving. Flat-band hosted call control at $99, $219, $399 and $699 CAD per month wins mainly by making growth free inside a band and exit close to free at any time [5]. That second property is worth more than the monthly difference, because it is what you will still have in year four.