Automation on a business line has three generations behind it and buyers are usually thinking of the wrong one. A menu that asks you to press two is not what is being sold now, and neither is a recording. What is sold now understands speech and holds a conversation, and the useful way to think about it is as a filter with a defined overflow rather than as a replacement receptionist.
A menu, a bot and a conversation are three products
A phone tree routes by keypress and resolves nothing; callers tolerate it and resent it. A scripted voice bot recognises a limited set of phrases and fails outside them, usually by repeating itself. A conversational system handles open speech, holds context across turns and can act on what it hears. They are priced differently and they fail differently, and a provider describing all three as automated answering is not being helpful. Ask which one you are being sold and ask to hear a real call rather than a demonstration.
Treat it as a filter with a stated pass rate
The honest measure is what share of calls it resolves without a human, and any provider running this at scale knows the number. If it resolves seventy per cent of calls, the human tier behind it only has to handle thirty, which changes what you need to buy there and is where the saving actually comes from. If it resolves twenty per cent, it is an expensive front door. Ask for the figure on accounts like yours rather than in general, and treat an unwillingness to give one as the answer.
The overflow decides whether it works
Every filter needs somewhere for the calls it cannot handle to go, and that somewhere is what you are really buying. Ask what triggers the handoff, whether the caller has to ask for it, how long it takes, and crucially whether a human is staffed at the hours you need. A system that hands off to voicemail overnight is an answering machine with a better front end, and it should be compared against one rather than against a receptionist service.
Write the flow before you buy the system
Automation removes the tolerance for a vague call flow. A human operator improvises around a gap; a system does the wrong thing confidently and repeats it identically for every caller. The businesses that get value here are the ones that already knew their call flow well enough to write it down. If buying automation is the way you are hoping to avoid that work, the result is usually worse than an operator who at least sounds uncertain when they are unsure.
Questions people ask about automated receptionist system
Is an automated receptionist just a phone tree?
No, though the words cover both. A tree routes by keypress and resolves nothing; a conversational system handles open speech and can act on it. Ask which you are being sold and ask to hear a real call rather than a demonstration.
How do I judge an automated receptionist system?
By the share of calls it resolves without a human, on accounts like yours. At seventy per cent it changes what you need to buy behind it; at twenty per cent it is an expensive front door.
What happens to calls it cannot handle?
That overflow is what you are actually buying. Ask what triggers the handoff, whether the caller must ask, how long it takes, and whether a person is staffed at your hours. Handing off to overnight voicemail makes it an answering machine.
Do I still need a written call flow?
More than before. A human improvises around a gap in a script; an automated system does the wrong thing confidently and identically for every caller. Write the flow first, or the automation will scale the mistake.