Is an AI receptionist worth it? Work it out properly

Everyone selling one of these has a calculator on their website, and every calculator returns yes.

Here’s how to work it out for yourself, including the four situations where the answer is a clear no. Most of the difficulty is that the deciding number is one almost nobody has actually counted.

The number that decides it

Not the monthly fee. Not a receptionist’s salary. What you currently lose to calls that go unanswered.

That figure is invisible by definition. A missed call leaves no trace, no message and no record. The caller rang the next business on the list and you never learned they existed. Which is precisely why people underestimate it, sometimes by a lot.

So before comparing any products, spend two weeks counting.

Counting it, properly

Missed calls per week. Your phone system or mobile already logs these. Count them for a fortnight rather than guessing, because guessing reliably produces a number that’s too low.

Subtract the noise. Sales calls, wrong numbers, the same person ringing three times in ten minutes. You want genuine enquiries only.

What proportion would have become work? Your own conversion rate on calls you do answer is the honest starting point. If half the people you speak to become customers, use half.

Average job value. Not your best job. The ordinary one.

Multiply those together and annualise it. That’s the size of the hole.

For a trade business missing five real enquiries a week, converting half of them, on jobs worth a few hundred dollars each, the annual figure lands somewhere that ends the discussion quickly. For a consultancy fielding two enquiries a month, it doesn’t. Same technology, opposite answer, and the only thing that changed was the arithmetic.

The other side of the ledger

Three costs, and you should ask for all three separately.

The monthly platform fee. Usually modest and usually quoted prominently.

Per-minute usage. Less prominent. Ask what a typical call costs and multiply by your actual volume, including the calls that turn out to be nuisance calls, because those get answered and charged too.

Setup. Anything that needs connecting to your calendar, your CRM or your phone system is work, and it’s where quotes diverge most. A proposal with no setup line is usually describing a system that answers and takes messages. That may be fine, but it’s not the same product as one that books jobs.

Worth being clear about the two ends of that. An off-the-shelf product you configure yourself is a monthly subscription and an afternoon. A system properly wired into your diary, your customer records and your payments is a build, and my own rates for that work are published here. Which one you need depends entirely on the count above.

Then the cost nobody quotes: your time maintaining it. Prices change, you add a service, your hours shift over Christmas. Something has to update the material the system answers from, and if nothing does, it will keep confidently quoting last year’s rates. Budget an hour every few months and decide now whose hour it is.

Four times the answer is no

You already answer nearly everything. If someone is reliably on the phone, the gain is small and you’re adding a system to maintain for very little return.

Your callers are usually upset. Businesses whose inbound calls skew towards complaints, disputes or distress should think hard. Handing those to an automated voice is a poor first impression at the exact moment it matters most.

Your enquiries need real consultation. If nothing can be quoted or booked without a proper conversation, the system can only ever take a message. Voicemail does that for free.

Your volume is genuinely low. A handful of calls a month rarely justifies a monthly fee, however good the product is.

None of these are technology failures. They’re just businesses where the number doesn’t work, and it’s better to find that out with a spreadsheet than a contract.

The thing that changes the answer

If your enquiries mostly arrive outside business hours, the calculation shifts sharply.

Calls at seven in the evening or on a Sunday morning currently go nowhere at all. There’s no competing arrangement to improve on and no risk of degrading a service you already provide. For a lot of trade and service businesses, out-of-hours coverage alone carries the entire business case, and everything that happens during the day is a bonus.

Worth checking your call log for the time-of-day pattern before deciding. It surprises people.

Don’t buy the whole thing first

The strongest position is a narrow one.

Start with overflow and after-hours only, so it handles calls that currently go to voicemail and touches nothing that’s presently working. If it disappoints, you’ve lost a monthly fee and nothing else. If it performs, you widen it with evidence rather than optimism.

Whatever you do, keep the before figure. You counted missed calls for a fortnight to make this decision. Count them again after three months, or you’ll end up judging the result on whether it feels better, which is not the same thing as whether it worked. That principle applies well beyond phone systems, and it’s the last section of the process mapping template for exactly this reason.

So, is it worth it?

If you miss calls regularly and your jobs are worth a few hundred dollars or more, almost certainly. That’s an unusually clear-cut case in a field where most answers are “it depends”, and it’s worth saying plainly.

If you’re in one of the four situations above, no, and you’ll save yourself a subscription by working that out now.

For everything in between it comes down to the count, which takes a fortnight and costs nothing.

Before that, it’s worth knowing what these systems can and can’t do and how they actually work, because both affect which column your business lands in.

If you’d rather talk it through, tell me roughly how many calls you’re missing and I’ll tell you honestly whether the numbers justify it. Sometimes they don’t, and that’s a perfectly good outcome for an afternoon’s thinking.