Should a small business use an AI receptionist?

The strongest case is a business whose phone rings while everybody is working. A trade on a job, a clinic with patients in the room, a shop with a queue at the counter.

The weakest case is a business where the call is the service. A consultant whose first conversation is the sale gains very little.

Most businesses are somewhere between, which is why the useful exercise is looking at the last fifty calls rather than reading a page like this.

It works when calls are routine and repetitive. Booking, hours, location, whether you cover a particular area.

It works when the alternative is voicemail. Most people will not leave a message, so the comparison is against nothing rather than against a person.

It works out of hours, where the choice is an assistant or a missed call until morning.

It works poorly for distressed callers, complex negotiation, or anything where being heard by a person is the point.

It works poorly with a thin brief. An assistant told very little hands over constantly and everybody blames the technology.

Say you own a bridal alterations shop and you sort your last fifty calls into piles. Twenty-two wanted to book a fitting. Nine asked, "Is my dress ready?" Six wanted hours or directions, and five asked about prices. Four were long, anxious conversations about a rush job before a wedding. The last four were junk.

Bookings, hours and price questions make 33 calls an assistant can finish by itself, since you can write every answer down. Those four anxious brides need you, and a quick transfer gets them to you.

That leaves the nine status calls, and they're the real design question. Your rack and your work tickets are invisible to the assistant, so it can't say whether a dress is ready. It can take the name and promise a callback, which beats voicemail but is slower than a person glancing at the rail. If status calls were half your volume, you'd want to fix that first, perhaps with a text to customers when work is finished.

Every business carries two risks on the phone: the call nobody answers and the call that's handled poorly. An assistant reduces the first and adds a little of the second. Whether that trade is a good one depends on what each costs you.

For the alterations shop, a missed call is a $300 job that goes to the seamstress across town. A clumsy call is a nervous bride who gets a slightly stiff booking experience and still turns up. Answering clearly wins that trade.

Change the business and it flips. For a hospice intake line or a bail bondsman, one mishandled call can do lasting harm, and missed calls are rare because someone is always on duty. There the math says no, or says after-hours overflow only.

Timing changes the answer as well. If most of your calls arrive while you're free to pick up, there's little to recover. The strongest case is a phone that rings during appointments or after closing, and your call log will show whether that's you.

GreetKeeper starts at $49 a month for 200 answered calls, with booking included on every tier rather than gated to a higher one.

A trained human backstop is a different product at a different price, which is what staffed services like Ruby and Smith.ai sell. We are the software kind, and we say so rather than blur the two.

Rosie is cheaper at entry, at $49 for 250 minutes, and it keeps booking behind its $149 plan. Both facts are worth knowing.