What phone automation really costs: market prices and a 5-step ROI worksheet

Nobody in this industry prints prices. Every pricing page hides behind a contact form, every case study leads with a percentage from someone else's business, and you are left wondering whether the thing costs 50 euros a month or 5,000. We build phone automation for small businesses, and the secrecy is, in our view, the main reason owners stall. So this post does the opposite: market prices for every option, the costs that never make the brochure, and a worksheet you can fill in before you talk to us or anyone else.

What are your four options for covering the phone?

When a customer calls while you are on a job, in a treatment room, or asleep, one of four things is covering that call. Here they are side by side, at typical market rates. None of these numbers is a quote. They are ranges, and your region and call volume move them.

Option 1: doing nothing is the expensive one

A study by 411 Locals found that 62 percent of calls to small businesses go unanswered. Not after hours. Overall. Industry consumer research adds the second half of the problem: roughly 85 percent of callers who hit voicemail do not leave a message, and most never call back. Put those together and the default option, changing nothing, means most of the demand ringing your phone evaporates without a trace. It never shows up in any report, which is exactly how it survives year after year.

Option 2: the answering service

An external service picks up with your company name and writes down a message. That beats voicemail. But a message is not an appointment, and speed decides whether the lead is still yours when you call back. Harvard Business Review's lead-response research found that reaching a lead within five minutes instead of thirty makes you about 100 times more likely to connect. A note saying a Mr. Weber called about a leaking pipe ages badly while you finish the job you are standing on.

Option 3: reception staff

A human at the front desk is still the best experience for a walk-in-heavy business, and nothing here argues against staff you need anyway. As pure phone coverage, though, it is the heavyweight option: in Germany a receptionist costs the employer roughly 2,500 to 3,500 euros a month once employer social contributions sit on top of gross pay. That buys about 40 of the 168 hours in a week. Evenings, weekends, lunch breaks, and sick days still fall through.

Option 4: the AI voice agent

The agent answers on the second ring, every time. It asks the questions you told it to ask, books the appointment into your calendar, and forwards anything unusual to you. Market prices vary with call volume and how much gets customized, but managed setups for a small business typically land between 200 and 500 euros a month, plus a setup fee. That is the option we sell, so weigh our bias accordingly. It is also why the rest of this post is a worksheet instead of a pitch.

The fastest way to judge option 4 is not reading about it. Call the live demo and try to book an appointment. Two minutes, no form, nobody rings you back.

Does it pay for itself? The five-minute worksheet

The SBE Council reported in 2026 that 74 percent of small-business owners say they need clearer ROI evidence before adopting AI tools. Fair. But the evidence you need is not in anyone's case study. It is in your phone log. Five steps:

1. Count your missed calls per week. Open the call log and count one ordinary week. Do not guess, because owners guess low. Worked example: 12 missed calls, of which 10 are genuine enquiries and 2 are spam.
2. Apply your booking rate. Of the enquiries you do answer, how many become jobs? Example: 30 percent. So 10 missed enquiries were worth 3 jobs.
3. Multiply by your average job value. Example: 400 euros per job. 3 jobs make 1,200 euros of missed revenue per week.
4. Take your margin, not your revenue. At a 40 percent margin, that is 480 euros of lost profit per week, roughly 1,900 euros a month.
5. Compare with the monthly cost. Take a market-rate figure, say 300 euros. Recovering 1,900 euros against 300 spent is roughly a 6x return, and the fee is earned back in the first week of the month. Even at the top of the market range, 500 euros, this example clears with room to spare.

Now run it with your own numbers, and be pessimistic on purpose. Halve the booking rate if you are unsure. If the result still clears the fee, you have your answer. If it only clears with optimistic inputs, treat that as a no.

Want a feel for step 2 before you trust it? Talk to the demo agent and watch it handle a booking. What it does with your call is what it would do with your customers' calls.

Which numbers are not in the brochure?

Some real costs rarely get mentioned. Count them in before you decide.

• Your own setup time. The vendor builds the agent, but the knowledge comes from you: services, prices, opening hours, what counts as an emergency, which calls to forward. Budget a few hours of your own time in the first week, not zero.
• The first two weeks are a tuning period. The agent will mispronounce a street name, misclassify a caller, or ask one question too many. A good provider reviews transcripts with you and fixes it fast, but expect corrections in week one, not perfection.
• Edge cases still need a human. A furious customer, a complicated complaint, a caller who will not state their business: those should be forwarded to you, not handled by a machine. If a vendor promises 100 percent automation with no human fallback, walk away.

When does the math not clear?

Sometimes it just does not, and you should hear that from a vendor at least once. Cases where we tell people not to buy:

• Low call volume. If you miss two or three calls a week, run the worksheet anyway: at a 400-euro job value the result lands near the fee, not comfortably above it. Below that volume, forget it.
• Low job value. If a typical transaction is 20 euros, a missed call costs you cents in margin. The math needs jobs worth booking.
• You already answer nearly everything. A solo consultant at a desk who picks up 95 percent of calls has no leak to plug. The worksheet says so at step 1, and no tool changes that.

If one of those is you, keep your money. Automation earns its fee where calls are being lost, nowhere else.

Not sure which side of the line you are on? Try the demo call first, then run the worksheet with your real numbers. If the math says no, you have lost two minutes.

FAQ

Is there a contract or lock-in?

Depends on the provider, so ask before you sign. The market has everything from month-to-month to 12-month terms. Our advice, and it applies to any vendor including us: do not accept a long lock-in before the agent has proven itself on your real calls. A provider who is confident in the product can live with monthly notice after the setup phase.

What if it makes mistakes?

It will, especially in the first two weeks. The realistic goal is not zero mistakes. A human receptionist does not manage that either. The realistic goal is that every call gets answered, unclear calls go to a human, and every conversation leaves a transcript, so mistakes get found and fixed instead of repeating. Ask any provider to show you how corrections work before you buy.

Do I need a new phone number?

No. Your published number stays. Calls reach the agent through a forwarding rule you control: after five rings, outside opening hours, or always. Turning it off is the same setting in reverse.

How fast is the payback, typically?

Whatever the worksheet says, nothing else. In the worked example above, the recovered margin covers a market-rate monthly fee within the first week. With low call volume it can take months or never happen, which is why the worksheet comes before any purchase. Anyone who promises a payback period without asking about your missed calls and your job value is reading from a brochure.