What to do about missed calls while you're actually on a job

This page may contain affiliate links from which Blur Studio earns commissions.

A missed call on a job site isn't just an inconvenience, it's usually a lead that goes to whichever competitor picks up first. Voicemail doesn't fix this; most callers hang up rather than leave a message, and the ones who do still wait for a callback that competes with the rest of your day. There are two real ways to close that gap: a human answering service, or an AI receptionist.

What does ElevenLabs' AI receptionist actually cost and do?

ElevenLabs sells its AI receptionist as its own product, Reception (also marketed as Reception by ElevenAgents), with pricing that's entirely separate from ElevenLabs' main Free/Starter/Creator/Pro/Scale/Business plans, which is a real distinction the product's own marketing doesn't make obvious at a glance. Reception's own tiers: Basic at $29 a month for 75 call minutes and one phone number, Plus at $79 a month for 275 minutes and three numbers, Premium at $199 a month for 1,000 minutes and five numbers. Go over your monthly minutes and you pay $0.30 to $0.45 a minute in overage, depending on which tier you're on.

What it actually does once it's set up: answers inbound calls around the clock in English, Spanish and more than 90 other languages and accents, checks your staff's availability and books appointments straight into Google Calendar, answers questions from a knowledge base you set up, and transfers to a real person when the call needs one. US-based businesses get a dedicated phone number during setup, and you can either forward your existing business line to it or hand the new number out directly; international setup needs some extra configuration.

A worked example on the cost side: a solo tradesperson missing 15 calls a week, averaging 2 minutes each if they'd been answered, is about 120 minutes a month, which fits inside Basic's 75-minute allowance only if some of those calls are shorter than average; realistically, most one-person operations land in Plus territory at $79 a month once you account for a few longer calls and some overflow. Compare that $79 against what even one of those calls is actually worth: a single missed job that would have been worth $300 or more makes the whole month's subscription pay for itself several times over if the receptionist saves even one job a month that would otherwise have gone to voicemail and been lost. That's the actual math worth running before deciding a monthly subscription "isn't worth it" for a small operation; the comparison isn't against doing nothing; it's against the value of the calls you're currently losing.

What I'd actually tell you if this is your situation

If most of your missed calls are simple, "are you available," "how much for X," "can you fit me in this week," an AI receptionist handles that volume well and consistently, at a fraction of what a human answering service costs per month. Where I'd push back on it: if your calls are frequently complex, emotionally charged, or need real judgment (an upset customer, a safety issue, a quote that depends on details only you can evaluate properly), a voice bot answering those calls can make a bad situation worse by sounding competent while actually missing the nuance a human would catch. Look at your actual call volume and call type honestly before picking either option; most businesses are a mix, and the right setup usually routes the simple stuff to AI and keeps a clear, fast path to a human for anything that isn't.

Step by step: deciding and setting this up

1. Track a week of missed calls (or check your phone's own call log) and sort them roughly by type: scheduling, pricing questions, emergencies, general questions.
2. Pick the Reception tier that covers your realistic monthly call-minute volume, not your lightest month; overage adds up fast if you underestimate.
3. Decide whether you want a dedicated new number or to forward your existing line; forwarding keeps your current number intact for customers who already have it saved.
4. Build out the knowledge base answers for your most common questions before going live, since the receptionist is only as good as what it's been told to say.
5. Set a clear transfer rule for anything that shouldn't be handled by AI, and test it with a real call before you trust it with real customers.

Compare Reception's own pricing against your current missed-call cost on ElevenLabs' site before committing to a tier.

Building a knowledge base that actually works

The receptionist is only as good as what it's been told to say, which means the setup work that actually matters happens before you go live, not after. Start with the five questions your office or your voicemail actually gets asked most often: your service area, your typical response time or next available slot, your call-out or diagnostic fee if you charge one, what you do and don't service, and how to get an emergency handled outside normal hours. Write the answers the way you'd actually say them out loud, not as a formal policy document; the receptionist reads from what you give it, so stiff, legalistic answers come out sounding stiff and legalistic on the phone too. Update the knowledge base whenever something genuinely changes (a price increase, a new service area, a seasonal availability change), since a receptionist confidently repeating outdated information is worse for trust than voicemail ever was.

Test it yourself before any real customer does. Call the number, ask it the five questions above plus a couple of edge cases (a service you don't offer, a location outside your area), and listen for where it handles the question well versus where it either makes something up or gets stuck. Fix those gaps in the knowledge base directly rather than assuming they'll sort themselves out with more calls.

The human answering service alternative

A human answering service costs more per month than an AI receptionist in most cases, but handles ambiguity, emotion and judgment calls better, at least for now. It's the safer default if your call volume is low enough that the cost difference doesn't matter much, or if your business genuinely depends on every caller feeling like they reached a real person who understood them, not a system, however good the system sounds. The cost and quality gap is laid out in AI receptionist vs human receptionist.

When should you skip this?

If your call volume is genuinely low (a handful of calls a week), the monthly cost of either option, human or AI, may not be worth it yet compared to just calling people back promptly. And if your business's calls are consistently complex or sensitive, don't let an AI receptionist be the first voice a customer in a real bind hears; route those straight to a person. A middle path worth considering before ruling either option out entirely: use the AI receptionist only outside your normal working hours, when the alternative is genuinely voicemail anyway, and keep answering calls yourself during the day when you're actually reachable between jobs. That gets you the after-hours coverage without changing anything about how your business runs during the hours you're already available.

How I'd approach this

Reception's own setup covers most small businesses well out of the box. Where I get involved is when someone wants it doing more than answering and booking: pulling real-time job status from your own scheduling software before it talks to a caller, routing by job type into different knowledge bases, or logging every call straight into your CRM with the outcome tagged automatically. That's a real integration on top of what Reception does by default, built around your actual systems rather than its generic setup.

A concrete version of that: a field service business running Jobber or Housecall Pro wants the receptionist to actually see the crew's real-time schedule before offering an appointment slot, not just a generic "someone will call you back" promise. That means the receptionist's booking flow needs to query the job platform's API in real time rather than relying on a static calendar that drifts out of sync the moment a technician runs long on a job. The payoff is a receptionist that can genuinely say "the earliest I can get someone out is Thursday at 2pm" rather than booking a slot that gets double-booked an hour later, which is the difference between a receptionist customers trust and one that quietly creates scheduling headaches nobody notices until a crew shows up to a conflict.