AI Receptionist for Solo Entrepreneurs: How to Test the ROI in 30 Days
Nobody can promise a solo operator 30-day payback, but you can measure it. Here is the break-even math, a day-by-day test plan, what each vendor's trial covers, and when it won't pay.
For a one-person shop, an AI receptionist pays for itself in a month only if it wins back more jobs than its monthly price divided by your profit on one job, which at an assumed $100 profit is 1.5 jobs for a $149 plan. No vendor can promise that in advance, but you can measure it in 30 days with your own call log and a trial. This guide gives you the break-even math, a day-by-day test plan, what each vendor's trial covers and the cases where it won't pay.
Disclosure: TheKeyBot, which publishes this guide, makes an AI receptionist. Our plans appear below next to other vendors' published prices and trials, and we use only what our own pricing and feature pages say about them.
Key takeaways
- Break-even is one division. Monthly cost divided by profit per job gives the jobs it must win back. At an assumed $100 profit, that is half a job for Rosie's $49 plan, 1.5 jobs for TheKeyBot Lite at $149 and 5 jobs for TheKeyBot Core at $500.
- Most free periods are shorter than 30 days. Rosie and Dialzara offer 7-day trials and TheKeyBot offers 14 days, so plan to pay for part of the test month.
- Measure against a baseline. Count last month's missed calls and how many of your callbacks connected before you switch anything on.
- Callbacks are the hidden leak. In a 2020 Pew Research Center survey, only 19% of U.S. adults said they generally answer calls from unknown numbers.
- It may not pay if most callers are repeat customers who wait for you, or if you're already booked solid.
What does "ROI in 30 days" mean for a one-person shop?
It means the jobs the AI receptionist wins back in a month earn more profit than the month costs. Revenue isn't the right measure, because a $160 lockout that costs you $60 in fuel, parts and time is worth $100 to the business, not $160.
Jobs it must win back = monthly cost / profit per job
Here is that division for entry plans across the market, at three assumed profit levels. Prices are October 2026 list prices; your volume may need a bigger plan.
| Plan | Monthly price | What it includes | At $60 profit a job | At $100 | At $150 |
|---|---|---|---|---|---|
| Rosie Professional | $49 | 250 minutes | 0.8 | 0.5 | 0.3 |
| Goodcall Starter | $79 | 100 unique callers | 1.3 | 0.8 | 0.5 |
| Dialzara Business Pro | $99 | 220 minutes | 1.7 | 1.0 | 0.7 |
| PATLive 50 minutes (live) | $99 | 50 minutes, about 20 calls | 1.7 | 1.0 | 0.7 |
| TheKeyBot Lite (pricing) | $149 | 100 calls | 2.5 | 1.5 | 1.0 |
| Smith.ai AI Pro | $150 | 75 calls | 2.5 | 1.5 | 1.0 |
| Ruby 50 minutes (live) | $250 | 50 minutes | 4.2 | 2.5 | 1.7 |
| TheKeyBot Core (pricing) | $500 | 500 minutes, plus scheduling, dispatch and payments | 8.3 | 5.0 | 3.3 |
The bar is low for the cheap plans, which is exactly why the test matters. A plan that costs half a job a month is easy to justify, but only if it wins back real jobs rather than taking messages you'd have caught anyway.
Where does a solo operator lose calls today?
When you're the only person, calls go missing at four predictable moments: while you're on a job, while you're driving, while you're asleep and while you're on another call. Each one has a different fix. A tech under a dashboard can't answer, a driver shouldn't, and a sleeping owner may answer half-awake and quote the wrong price.
You don't need an industry figure for how many you miss. You need yours, and you can get it in an hour.
The baseline worksheet
Fill this in from the last 30 days before you turn anything on. The example column is made up to show the math.
| Line | What to count | Where it comes from | Example (assumption) |
|---|---|---|---|
| A | Calls, minus spam and repeat dials | Phone or carrier log | 75 |
| B | Answered by you, live | Log outcome | 50 |
| C | Missed | A minus B | 25 |
| D | Missed callers who left a voicemail | Voicemail box | 9 |
| E | Callbacks that reached the caller | Outbound log | 8 |
| F | Jobs booked from missed calls | Job board | 4 |
| G | Average ticket | Invoices | $160 |
| H | Direct cost per job (fuel, parts, your time on site) | Invoices and receipts | $60 |
| I | Profit per job | G minus H | $100 |
Line E is often the surprise. In Pew Research Center's July 2020 survey of 10,211 U.S. adults, 67% said their general practice is not to answer unknown numbers but to check voicemail if one is left. If you call back from your personal cell, you're the unknown number. Our guide to why lockout callers hang up on voicemail shows how to track it.
The 30-day test plan
Decide what "pass" means before day one, so the result can't talk you into anything. A simple rule: keep it if the jobs won back cover the cost with room to spare, such as twice the break-even number from the table above.
| Days | What to do | What to record |
|---|---|---|
| Day 0 | Fill in the baseline worksheet; write down your pass rule | Lines A to I |
| Days 1-2 | Load your business name, hours, service area and what to say about price; forward only the calls you miss, or nights only | The forwarding setup you chose |
| Day 3 | Place test calls: a late-night lockout, a price shopper, a Spanish-speaking caller, a request for a person, an out-of-area address | What each call produced and how fast it reached you |
| Days 4-27 | Run it live and return every message as fast as you can | For each AI-answered call: time to your callback, reached or not, booked or not |
| Days 28-30 | Rebuild the worksheet for the test month and apply your pass rule | Jobs won back, total cost including any overage |
Two setup choices matter more than the vendor:
- Forward only what you miss. Many phone systems can forward unanswered calls after a set number of rings; Verizon Business Digital Voice, for example, lets you pick the number of rings for its "When no answer" forward. That way you keep answering the calls you can take, and the AI handles the ones you would have lost, which is what the test should measure.
- Return messages fast. A message-taking plan only wins the job if you call back while the customer is still waiting. Set yourself a target, such as 15 minutes, and log how often you hit it.
Our trial test-call script has the full list of calls to place on day 3.
What should the AI collect on every call?
A message only wins the job if you can act on it without calling back to ask basic questions. Before day 1, write down what every message must contain, and fail any test call that leaves something out.
| Field | Why a solo operator needs it |
|---|---|
| Caller's name and the number to call back | So your callback reaches the right person on the first try |
| Exact location, or cross streets and a landmark | So you can quote travel and give an arrival window |
| What's locked or lost: car, house, business, safe | So you know which tools and stock to bring |
| For cars: year, make and model, and whether any working key exists | So you can price the job before you call |
| Safety check: is a child, a vulnerable adult or a pet locked inside? | Those callers should be told to call 911 first |
| How urgent it is, and when they can meet you | So you can order your callbacks |
| How they found you | So you know which ads and listings are working |
Our guide to identifying the key type on the phone covers what to ask when the caller doesn't know. On Lite, our Lite page says callers reporting a child or pet locked in a car are told to call 911 first; check that any vendor you test does the same.
Which trials cover the test?
Free periods vary, and few cover a full month. This is what each vendor publishes as of October 2026.
| Vendor | Free period or allowance | After it ends |
|---|---|---|
| TheKeyBot (pricing) | 14-day free trial on Core, Pro and Elite; Lite's first 5 answered calls are free, up to 7 days (Lite) | Lite $149 a month; Core $500 |
| Smith.ai AI Receptionist | A $0 plan with 25 calls a month | $3.00 a call after 25 |
| Rosie | 7-day free trial | From $49 a month; payments are non-refundable per its terms |
| Dialzara | 7-day free trial on inbound voice plans | From $29 a month |
| Goodcall | A free demo agent instead of a trial | From $79 a month |
| PATLive (live) | 14-day free trial on any plan | From $49 a month plus $2.99 a minute |
| Posh (live) | 7 days or 250 receptionist minutes, whichever comes first | From $65 a month plus $2.30 a minute |
| Ruby (live) | Money-back guarantee for first-time users within 21 days or 500 minutes, whichever comes first | From $250 a month |
So the 30-day test usually has a free part and a paid part. If your pass rule fails at day 30, cancel by the vendor's rules; Smith.ai's AI plan, for example, asks for 30 days of notice.
How do you know a job was really won back?
This is where most ROI claims go wrong, so be strict. Count a job as won back only if all three are true:
- The call came when you couldn't answer, on a job, driving, asleep or on another call. If you would have picked it up anyway, the AI didn't win it.
- The AI took the call and passed you what you needed to act: name, number, location and the job.
- You booked the job from that message or booking, and it wasn't a repeat customer who would have waited for your callback regardless.
Then compare the test month's count with line F of your baseline. The difference, not the total, is what the AI won back.
A worked example, both ways
Every number here is made up to show the math. The same shop runs the test twice, once with mostly new callers from search and once with mostly repeat customers.
| Line | Mostly new callers | Mostly repeat customers |
|---|---|---|
| Missed-window calls the AI answered | 25 | 25 |
| Messages you returned within 15 minutes | 22 | 22 |
| Jobs booked from AI-answered calls | 9 | 5 |
| Jobs booked from missed calls in the baseline month (line F) | 4 | 4 |
| Jobs won back | 5 | 1 |
| Profit won back at $100 a job | $500 | $100 |
| Cost of a $149 plan | $149 | $149 |
| Net for the month | +$351 | -$49 |
The second column isn't a failure of the software. In this example, repeat customers who knew the owner waited for the callback, so there was little to win back. That's the honest case for staying with voicemail, and the test is how you find out which shop you are.
When won't an AI receptionist pay back for a solo operator?
- Most of your calls are repeat or referral customers who will wait for you, as in the second column above.
- You're already booked solid. Winning back a job you can't fit in doesn't add profit; it adds a customer you have to disappoint.
- Your callers need you personally, such as long-standing commercial accounts that call your cell.
- Your callers speak Spanish and the plan doesn't. In the Census Bureau's 2018-2022 survey, 61.1% of U.S. residents who spoke a language other than English at home spoke Spanish, and Goodcall says its agent speaks English only for now.
- You can't return messages quickly. A message-taking plan depends on your callback.
If two or more of these describe you, run the test with a cheap plan or keep your current setup and recheck in six months.
Does a solo shop need message-taking or booking?
It depends on what happens between the message and your callback.
Message-taking suits an owner who wants the final say on every price and can call back within minutes most of the time. TheKeyBot Lite works this way: $149 a month with 100 calls included, then 50 cents a minute, answering 24/7 in English and Spanish and sending the message to your phone. Our Lite page says it doesn't quote, on purpose, and that callers who hang up before leaving details get a text asking what they need. Our Lite explainer covers the setup.
Quoting and booking suit an owner who often can't call back for an hour, because the customer gets a price and a slot on the first call. Among vendors, Rosie's Scale plan, Dialzara, Smith.ai and Goodcall describe calendar booking. TheKeyBot Core ($500 a month for 500 minutes, then 45 cents) adds scheduling, GPS dispatch, POS and payments, and our quoting page describes quotes from a price sheet you upload. At $500, Core needs five jobs a month at a $100 profit, so it fits a solo shop only when it also replaces other software. Our Lite vs Core guide shows where that line falls.
Do missed calls cost you more than the job?
They can, if you advertise. Google's Local Services Ads help is direct about it:
"Missed calls may negatively affect your responsiveness."
That line is from Google's ad ranking page, which counts responsiveness among the factors that feed the ad auction. The default ad schedule runs 24 hours a day, 7 days a week. A solo operator who keeps that default and misses night calls is advertising at hours when nobody answers. Either cover those hours or change the schedule; our guide to Local Services Ads and answer rate explains the settings.
What to check on day 30
| Check | Pass looks like |
|---|---|
| Jobs won back (strict count) | At least your pass rule, such as twice break-even |
| Total cost, including any overage | Matches what you budgeted from the pricing page |
| Messages returned within your target | Most of them, or the plan can't work for you |
| Test calls | Usable message every time, right language, no invented prices |
| Your time | Fewer interrupted jobs and no unsafe calls while driving |
If the shop passes, keep the plan and recheck the numbers every quarter. If it fails, cancel and keep the worksheet; it tells you what to fix first, whether that's callback speed, your ad schedule or simply that your customers wait for you. For a two-truck version of this math, see our two-truck ROI guide, and for answering safely on the road, our guide to taking calls while driving.
Frequently Asked Questions
Can an AI receptionist pay for itself in 30 days?
It can if it wins back more jobs than its monthly price divided by your profit per job, which is 1.5 jobs for a $149 plan at an assumed $100 profit. Nobody can promise that in advance, so measure it with a baseline month and a strict count of jobs won back.
How many calls does a solo operator need for an AI receptionist to make sense?
There's no fixed number, because the test is jobs won back, not calls. A shop with few calls but many missed new customers can pass, while a busy shop whose callers wait for the owner may not.
What is the cheapest AI receptionist for a one-person shop?
On October 2026 list prices, low entry points include a $0 plan with 25 calls a month at Smith.ai, $29 for 60 minutes at Dialzara and $49 for 250 minutes at Rosie. Check what each does on the call, because the cheapest plans mostly take messages.
Should I forward all my calls or only the ones I miss?
For the test, forward only the calls you miss or only nights. That keeps you answering the calls you can take and measures what the AI adds on the ones you would have lost.
How do I know a job was really won back by the AI?
Count it only if the call came when you couldn't answer, the AI passed you what you needed, and you booked it from that message. Subtract the jobs your callbacks already recovered in the baseline month.
How much does TheKeyBot Lite cost?
TheKeyBot Lite is $149 a month with 100 calls included, then 50 cents a minute, and your first 5 answered calls are free (up to 7 days). It answers 24/7 in English and Spanish and takes messages but does not quote or book; Core starts at $500 a month, and Core, Pro and Elite include a 14-day free trial.
Sources
- Pew Research Center, "Most Americans don't answer cellphone calls from unknown numbers" (December 14, 2020): https://www.pewresearch.org/short-reads/2020/12/14/most-americans-dont-answer-cellphone-calls-from-unknown-numbers/
- Google Local Services Help, "About ad rankings": https://support.google.com/localservices/answer/7527305?hl=en
- Google Local Services Help, "About ad scheduling": https://support.google.com/localservices/answer/9251526?hl=en
- Rosie pricing: https://heyrosie.com/pricing
- Rosie terms of service: https://heyrosie.com/legal/terms
- Dialzara pricing: https://dialzara.com/pricing
- Goodcall pricing: https://www.goodcall.com/pricing
- Goodcall FAQ: https://www.goodcall.com/faq
- Goodcall, how it works: https://www.goodcall.com/how-it-works
- Smith.ai AI Receptionist pricing: https://smith.ai/pricing/ai-receptionist
- PATLive pricing: https://www.patlive.com/pricing/
- Posh pricing: https://www.posh.com/pricing
- Ruby plans and pricing: https://www.ruby.com/plans-and-pricing/
- Ruby FAQs: https://www.ruby.com/faqs/
- U.S. Census Bureau, language spoken at home, 2018-2022 ACS (December 7, 2023): https://www.census.gov/newsroom/press-releases/2023/language-at-home-acs-5-year.html
- TheKeyBot pricing: https://www.thekeybot.com/pricing
About the Author
TheKeyBot Team is dedicated to helping locksmiths grow their businesses through AI automation and smart technology. With years of experience in the locksmith industry, our team provides actionable insights and proven strategies.
