Should I hire a receptionist or automate the phone?

Decide by what you are buying, not by what it costs. Hiring puts a person in your business who does a dozen things besides answer calls; an automated answering system does one thing and does it on every call, 24/7, for $0.60 CAD per answered minute with $0 setup and no contract. If your list of non-phone tasks fills half a page, hire. If the only thing on it is "pick up the phone", you were never hiring a receptionist — you were buying coverage, and a salary is an expensive way to buy it.

What are you actually choosing between?

Two different purchases that touch the same phone.

Hiring is a job. A receptionist greets the person at the counter, signs for the delivery, fixes the schedule when two appointments collide, and notices that the man in the waiting room has been sitting there forty minutes. Answering the phone is one line on that list, and rarely the longest one.

An automated system is a function. It answers the line, holds a spoken conversation, screens the caller, books into your calendar, and hands over to a person when your rules say so — on the first ring, on every call, at two in the morning. It does not sign for deliveries.

Compare the two on price and you get the wrong answer, because you are comparing a salary against a phone bill. The useful exercise is narrower: write down everything you want this person to do, cross off everything that is not a phone call, and look at what is left.

What does hiring a person commit you to?

More than the wage, and more than most first-time employers plan for.

Ontario's general minimum wage is $17.60 an hour, rising to $17.95 on 1 October 2026 — the legal floor rather than a front-desk rate, and it moves every year. On top of whatever you advertise sit employer CPP and EI contributions, vacation pay, and a WSIB premium if your industry requires one. The full employer arithmetic is worked through in virtual receptionist vs a full-time hire; this post is about the decision rather than the spreadsheet.

The commitments that never reach the spreadsheet decide how the hire feels in month six:

• Recruiting. Writing the posting, reading applications, interviewing, checking references.
• Training. Two to eight weeks before the person answers a call the way you would have answered it.
• Coverage. Vacation, sick days, appointments, the fortnight in July. Somebody covers the desk, and that somebody is usually you.
• Management. Scheduling, reviews, and the conversation when something goes wrong.
• Turnover. Front-desk roles turn over, and every departure restarts the first two items on this list.

Then the coverage arithmetic, which is the part that surprises people. A week holds 168 hours. A full-time hire covers about 40, fewer once lunch and leave come out. The other 128 are evenings, overnight, weekends, and statutory holidays. During the 40, one person answers one call at a time — so the second caller in a busy ten minutes hears ringing, and the third hears it longer.

What does automating the phone commit you to?

Three things, and the second is the one vendors tend not to mention.

You own the wording. The greeting, the way your services are named, which prices may be quoted by phone, your hours including the closures. A person picks that up by working beside you for a month. Software takes it from a list, and somebody has to write the list once, properly.

You own the escalation rules. Which callers go to a person immediately, whose phone rings, what happens when that phone is not picked up, what the agent does at midnight with nobody on call. A receptionist walks down the hall and interrupts you when something warrants it. Software does exactly what the rule says, which is a strength on routine calls and a liability when the rule is vague. Callers are told they are speaking with an AI by default; turning that off is your decision rather than ours.

You own the review, at least at the start. Reading flagged calls takes about ten minutes a day in week one, then roughly twenty minutes a week once the obvious gaps are closed. Skip it and every call is still answered — the agent simply stays as good as it was on launch day. That is a real ceiling, worth naming before you buy rather than after.

What you are not committing to is duration. Setup is $0, there is no monthly minimum, forwarding your line takes about ten minutes at your carrier, and forwarding it back takes the same ten. Compare that with the notice period in an employment contract.

Which calls need a person, and which do not?

Sort last month's calls into three piles and the decision largely makes itself.

Pile one: repetition. Hours, directions, do-you-do-this, the price of a standard job, book me in, move my appointment, is my car ready. On most small-business lines this is the large majority of volume, and it rewards consistency rather than judgment. The two-hundredth caller gets the same wording as the fourth.

Pile two: judgment. The upset customer whose complaint is partly fair. The bereaved family. The caller whose real question sits three questions underneath the one they asked. A trained person is better at these and will stay better for some time, which is why the agent moves them to a human with everything said so far attached rather than pressing on through a script. Where each side wins is set out in AI receptionist vs a human receptionist.

Pile three: the desk. Walk-ins, deliveries, filing, the waiting room. Not a phone problem at all, and no phone system touches it.

If pile two is most of your volume, hire. If pile one is most of it and pile three is empty, you are buying coverage rather than a person. If pile three is long, you are hiring regardless — and the question becomes what covers the phone during the 128 hours that hire does not.

Hiring, automating, or both: how do they compare?

The cost line is the one that behaves differently. At $0.60 CAD per answered minute with $0 setup, 100 answered minutes a month is $60, 300 minutes is $180, and 600 minutes is $360. Billing runs by the second, so the robocalls and wrong numbers that make up a real share of any business line cost about twenty cents each rather than a full call charge. There is no per-seat or per-line fee, so the third simultaneous call costs the same per minute as the first — where one hire would have put that caller into voicemail.

If your phone log carries no durations, two minutes a call is a reasonable planning figure — an assumption rather than a measurement, to be replaced with your own average after a month of real transcripts. Fifty forwarded calls at two minutes is 100 answered minutes, so $60. The whole rate card fits in a paragraph on the pricing page.

When is hiring the right answer?

Four cases, and we would rather name them than win an argument that sours in month three.

You need a front desk. Somebody has to be physically present — greeting patients, taking payment at a counter, managing a waiting room. Hire, and do not let a per-minute rate talk you out of it.

Most of your calls are pile two. Where nearly every call is delicate, unusual, or valuable enough to warrant a person on it, put a person on it. Law firm intake often sits near this line, which is why the sensible arrangement there is screening by software and substance by a person.

You want a human voice on every single call, with no software anywhere in the chain. That is a legitimate preference, and a staffed service with named receptionists — Moneypenny, for instance — is the honest recommendation for it, at several times the per-minute rate.

The volume is genuinely large. Past a few thousand answered minutes a month you are staffing a small call centre, and this stops being a hire-or-automate question.

When is automating the phone not worth it?

Three situations, plainly, and they come up more often than a vendor will admit.

Your phone barely rings. If you take fifteen or twenty calls a month and answer nearly all of them yourself, the phone in your pocket is already doing the job. Automating it buys a tidier process and very little else.

Nobody in your business will write the list. The wording, the prices that may be quoted, the escalation rules — somebody has to sit down for an hour and decide them. If that hour is never going to happen, the agent launches on guesses, and a confident wrong answer on your line is worse than a voicemail. This is the most common reason a deployment disappoints, and it is not a software problem.

The call is the service. Some consultants, coaches, and practitioners are selling the first conversation itself. A screening layer in front of that relationship costs more than the missed calls do. A salon chair is the opposite case: the booking is administration and the service happens in person, so the phone is safe to hand over.

In all three cases, automation does not make the difficult call easier. It moves the difficult call to a person faster, with more written down. That is a genuine improvement and it is not the same thing as solving it.

How do you decide in one week?

Six steps. The first four take about an hour between them, and none involves signing anything.

1. Pull last month's call log. Your phone provider has it — call count, and duration where the log carries it. Not a figure from memory; owners consistently underestimate how many calls ring out while they are working.
2. Sort it into the three piles. Repetition, judgment, desk work. Count each. The decision is mostly in that column of numbers.
3. Write the non-phone job description. Everything you want this person doing that is not answering a call. If it fills half a page you are hiring; if it is empty you are buying coverage.
4. Price both annually. Employer cost for a year against answered minutes for a year — never an hourly rate against a per-minute rate.
5. Test it before you decide. Call it yourself as an ordinary customer, then ask something it cannot know and check that it says so rather than inventing an answer. Ours answers on +1 437-494-9110 and the agent is called Maryann.
6. Run it narrow for one month. Forward only after-hours and overflow calls. Nobody's day changes, and at month end you hold a transcript of every call that used to reach voicemail.

Then count how many were real. If thirty were genuine enquiries you have a hiring case built on evidence rather than a hunch. If three were, you have just saved yourself a salary.

What do most owners end up doing?

Both, in the cheaper order.

The arrangement that holds up is a person at the desk during business hours, doing the parts of the job that were never about the phone, and an automated line covering the other 128 hours plus the overflow when the desk is already busy. Nobody loses a job, and the calls that were reaching voicemail stop reaching voicemail. Coverage of that shape is priced by trade and city on pages like plumbing call answering in Hamilton, and works the same way on a law firm intake line.

If you have not hired yet, run the line first. A month of transcripts is a better hiring brief than any job posting written from memory, and it costs about the price of a tank of fuel to find out.

Ontario minimum wage figures are from the Ontario Ministry of Labour, Immigration, Training and Skills Development, confirmed 24 September 2026; the general rate changes annually on 1 October.