An AI receptionist that answers 24/7 starts at around £10–£149 a month in the UK. A human answering service with genuine 24/7 live cover starts at $44 a month plus $1.54 a minute, with bundled plans from $159 to $329. On published prices from the same vendor — Smith.ai sells both — the AI tier costs $2.00 a call against $10.00 for the human tier. For most service businesses that puts the break-even at under one recovered job a month.
The harder question is how many calls you're actually losing. And the statistics circulating on that are, on inspection, worse than useless.
- The "$126,000 a year in missed calls" figure repeated across dozens of vendor blogs is untraceable — one attribution points at a 404, another names a different company entirely. Ambs Call Center, most often named as the source, actually publishes $26,000+ a year and $12.15 per missed call.
- The "62% of calls go unanswered" figure is real but comes from a January 2016 study of 85 businesses over 30 days. At least one vendor blog re-dates it to 2024.
- Nobody has credible current data on what your missed calls cost. You can work it out from your own call logs in about twenty minutes, and the method is below.
- Verified entry pricing, retrieved 13 August 2026: AI receptionists from £10/mo (The VoIP Shop) to £149/mo (Hand On Web) in the UK; US human 24/7 services from $44/mo + $1.54/min (Specialty Answering Service) to $329/mo (Abby Connect).
- No UK human answering service publishes a complete rate card; three of six publish nothing at all. Half the US providers do. That opacity is itself a cost.
- Article 50 of the EU AI Act applied from 2 August 2026. The disclosure duty falls on your AI vendor, not on you — but it's your greeting, so check what it says.
What does a missed call actually cost?
A missed call costs you the gross margin on the work you would have won from it, multiplied by the probability you'd have won it. Nothing more exotic than that.
The formula:
For a plumbing firm with a £250 average job, 50% margin and a 40% close rate on inbound enquiries, a missed call costs £50. For a private dental practice with a £2,000 average treatment plan, 70% margin and a 30% close rate, the same missed call costs £420.
The spread between those two numbers is the reason no single industry-wide figure means anything for your business. A missed call at a nail salon and a missed call at a commercial roofing contractor differ by two orders of magnitude, and any statistic that averages them is arithmetic with no decision attached to it.
There's a second cost most calculations omit: the caller doesn't wait for you. They ring the next firm on the results page, and by the time you call back the job is usually gone.
The best evidence on that comes from outside the phone industry. In March 2011, James Oldroyd, Kristina McElheran and David Elkington published an audit of 2,241 US companies in Harvard Business Review, alongside an analysis of 1.25 million sales leads across 29 B2C and 13 B2B firms. Companies that made contact within an hour of an enquiry were nearly seven times as likely to qualify the lead as those that tried an hour later, and more than 60 times as likely as those who waited 24 hours or more. Twenty-three per cent of the audited companies never responded at all.
That study measured web-form leads, not phone calls, and it's fifteen years old — treat it as directional rather than a phone benchmark. But the direction is unambiguous, and a call that rings out at 7pm and gets returned at 9am the next morning sits squarely in the 24-hour bucket.
Can you trust the missed-call statistics you've read?
No. Almost every article on this subject rests on two statistics — that the average small business loses about $126,000 a year to missed calls, and that 62% of business calls go unanswered — and neither survives being traced to source. The $126,000 figure has no reachable origin. The 62% figure comes from a 30-day study of 85 businesses published in January 2016, which at least one vendor blog re-dates to 2024. Here's the trail on each.
The $126,000 figure is untraceable
Untraceable — do not citeAira's blog attributes it to Ambs Call Center and links to ambscallcenter.com/the-ambs-blog/how-much-do-missed-calls-cost-your-business. That URL returned a 404 when we checked on 13 August 2026.
Phone2's blog attributes the same $126,000 to a different company — Dialora — with no study name, sample size or methodology.
One number, two mutually exclusive origins, and the attribution that names a specific company points at a page that no longer exists.
What Ambs Call Center has actually published is a different figure. Their report, promoted by press release on 21 August 2025 and authored by company president Aaron Boatin, states: "A single missed call costs the average business $12.15. For small to medium-sized businesses, the total annual loss often exceeds $26,000." It gives $8,800+ a year for a business missing two calls a day, and $26,000+ for one missing around six.
We can't prove $126,000 is false — you can't disprove a number with no methodology behind it. What we can say is that it cannot be verified, that the company most often named as its source publishes $26,000, and that a figure you cannot trace is not a figure you should budget against.
Worth stating plainly: Ambs sells human answering services, so it has a commercial interest in missed calls looking expensive. Its stated methodology is "an industry-backed cost-per-call model and conservative revenue estimates" — a model, not a measurement, with no disclosed sample size. That makes it the most transparent number in the category, which tells you more about the category than about the number.
The 62% figure is real, and it's from January 2016
Real — but from January 2016This one has a genuine source, and it is older than almost everyone citing it seems to realise.
It comes from a 411 Locals study published 18 January 2016. The methodology, verbatim: "We monitored the phone calls of 85 businesses, operating in 58 industries, for a period of 30 days." The findings: 37.8% of calls answered, 37.8% to voicemail, 24.3% no response at all. Add the last two and you get the 62% that circulates.
Aira's blog dates that study to 2024. It is from 2016 — a decade old, thirty days long, eighty-five businesses, run by a marketing agency rather than a research firm. It predates the iPhone X. It predates every AI receptionist vendor in the tables below.
None of that makes it wrong. Small businesses probably do miss a lot of calls. But a thirty-day snapshot of 85 businesses from 2016, re-dated to 2024 and repeated across hundreds of pages, is not evidence of what your business is losing in 2026.
The practical upshot: don't build a business case on any of these numbers. Build it on your own call data, which you already have and which is a decade more current than the best study in the category.
How to work out your own number in about twenty minutes
Every VoIP system and most mobile carriers will export a call log. That log contains everything you need.
Export the last 90 days of inbound calls.
You want timestamp, duration and answered/unanswered status. In RingCentral, 8x8, Dialpad and 3CX this is under Analytics or Call Log; on a mobile-only setup, your carrier's business portal has it.
Filter to unanswered calls outside your opening hours.
This is the number the AI is actually competing for. Calls missed during office hours are a staffing problem, and an AI receptionist is a partial fix at best.
Deduplicate by number.
A caller who rings three times in ten minutes is one lost enquiry, not three. Skipping this step is how businesses talk themselves into a number two or three times too high.
Count distinct after-hours callers per month.
Call this N.
Multiply.
N × average job value × gross margin % × close rate. Use your actual close rate on answered enquiries from your CRM or job book — not an industry benchmark.
A worked example, using illustrative figures for a small electrical contractor. Substitute your own — the point is the method, not these numbers:
Against that, one converted job returns £340 × 55% = £187 in gross margin. A £99/month AI receptionist therefore breaks even at 0.53 jobs a month: one recovered caller every eight weeks covers the subscription, out of the 22 it answers each month.
That's the whole business case. It turns on N, not on the subscription price, which is why the ROI question is usually less interesting than vendors make it — and N is the one number no blog post can tell you.
What does an AI receptionist cost in 2026?
Entry pricing, taken from each vendor's own live pricing page on 13 August 2026. Prices exclude VAT where not stated; no UK vendor on this list publishes VAT treatment.
UK vendors
| Vendor | Entry plan | Included | Overage | 24/7 | Setup fee |
|---|---|---|---|---|---|
| The VoIP Shop | £10/mo | Not stated at entry tier | £0.85 per call | Yes | None stated |
| Moneypenny | "From £25/mo" | Not published | Not published | Not stated | Quote-gated |
| Arrow | £99/mo | 150 minutes | Not published | Yes | Not stated |
| Hand On Web | £149/mo | 200 minutes | Not published | Yes | £199 |
The VoIP Shop is the outlier worth noting: at £0.85 per call it prices per call rather than per minute, which is unusual for AI and materially better if your callers are chatty. Their Premium tier is £30/month with 50 calls included, Ultimate £120 with 200.
US vendors (USD, for reference)
| Vendor | Entry plan | Included | Overage | Model |
|---|---|---|---|---|
| ai-receptionist.com | $14/mo | 60 min | $0.25/min | Per-minute |
| Dialzara | $29/mo | 60 min | $0.48/min | Per-minute |
| Rosie | $49/mo | 250 min | Not published | Per-minute |
| Goodcall | $79/mo | Unlimited minutes, 100 unique customers | $0.50 per extra customer | Per-customer |
| Frontdesk | $99/mo | 200 min | $0.25/min | Per-minute |
| Smith.ai AI | $0 (25 calls) / $150 Pro | 75–300 calls | $2.00/call (Pro) | Per-call |
| Slang.ai | From $399/location | Not published | Not published | Not published |
Two traps in that table. Frontdesk (formerly My AI Front Desk) advertises a $20 plan, but that tier includes zero voice minutes and does not answer calls — the real entry point for answering is $99. And Goodcall's "unlimited minutes" is true as stated: the meter is unique callers, not time, so a 40-minute call costs the same as a 40-second one, but your 101st distinct caller costs $0.50.
The short version: AI receptionist entry pricing clusters between £10 and £149 in the UK and $14 and $99 in the US, with 24/7 cover standard at every tier. What varies is the meter — per minute, per call, or per unique caller — and that choice matters more to your bill than the headline price.
What does a human answering service cost?
Same method, same date. Only providers with genuinely 24/7 live human cover are listed — Davinci Virtual's $129 entry plan is excluded because its live receptionists work 8am–8pm ET Monday to Friday, with automation outside that.
| Provider | Entry plan | Included | Overage | Setup fee |
|---|---|---|---|---|
| Specialty Answering Service — Economy | $44/mo | 0 min | $1.54/min | None |
| PATLive — Basic | $75/mo | 0 min | $2.60/min | None |
| Specialty Answering Service | $159/mo | 100 min | $1.44/min | None |
| PATLive — Starter | $250/mo | 75 min | $2.35/min | None |
| Ruby | $250/mo | 50 min | Not published | None |
| Smith.ai (human) | $300/mo | 30 calls | $11.50/call | None |
| AnswerConnect | $325/mo | 200 min | $1.95/min | $49.99 |
| Abby Connect | $329/mo | 100 min | Not published | Not published |
The two pay-as-you-go tiers at the top are the cheapest genuine 24/7 human cover on the market, and both carry a monthly base fee before a single minute is answered.
Ruby charges $250 a month for fifty minutes — five dollars a minute at the base rate — and publishes no overage rate anywhere on its site, though its terms confirm overage is charged and that calls are rounded up to the next 60 seconds. Abby Connect publishes per-chat overage but not per-minute. AnswerConnect's own website hides its rate card behind an email capture form; the figures above come from its publicly hosted PDF, linked in the table rather than the page you'd actually land on. That's three of eight rows where the published price required more than reading the pricing page.
The UK situation is worse
Not one UK human answering service publishes a complete rate card.
- Moneypenny UK, alldayPA and answer4u publish no price at all. Moneypenny's only published rate card is on its US site, in dollars.
- JAM publishes "prices starting at £35 a month" and nothing else.
- Verbatim publishes "from as little as £2.25 per day" and a £1/minute rate, with no package prices or minute allowances.
- Face for Business publishes "from £49 p/m" with no call or minute allowance attached to any tier, and its contract terms contradict each other across its own pages — the pricing page says "no long-term contracts", the small-business page says "we ask that you start on a 90-day minimum period".
Face for Business is also not a 24/7 service: 8am–8pm weekdays, shorter at weekends. Verbatim's core hours end at 6.30pm.
This asymmetry is the single most useful finding in the pricing research. Half the US providers publish an allowance-and-overage table you can budget from directly; Ruby, Abby Connect and AnswerConnect don't, or don't on the page you'd land on. Zero UK providers publish a complete one. If you're a UK business comparing options, you can price an AI receptionist off a web page in four minutes and you cannot price a human service without a sales call.
AI vs human answering service: which is actually cheaper?
Smith.ai is the cleanest comparison available, because it sells both and publishes both:
| Human tier | AI tier | |
|---|---|---|
| Entry price | $300/mo | $150/mo (Pro); $0 free tier |
| Calls included | 30 | 75 |
| Cost per call at entry | $10.00 | $2.00 |
| Overage per call | $11.50 | $2.00 |
| 24/7 | Yes | Yes |
Entry-tier published pricing, same vendor, retrieved 13 August 2026.
Same company, same brand promise, same category. The AI costs a fifth per call — $2.00 against $10.00.
That ratio holds broadly across the market. A UK business paying £99/month for 150 AI minutes is paying £0.66 a minute. The US human services that publish rates charge $1.44–$2.60 a minute. Those are different currencies in different markets and we're not going to pretend a conversion makes them comparable, but the gap is wide enough that the exchange rate isn't what decides it. As for the providers who publish nothing — they may well be cheaper. We have no way to know, and neither will you until you sit through the sales call.
Choose the human service if: your calls are emotionally loaded — bereavement services, healthcare triage, complaints escalation — legally sensitive, or need judgement that can't be scripted. That's the whole list, and note that price isn't on it.
We looked specifically for the low-volume case where a human pay-as-you-go plan undercuts an AI subscription, and it doesn't exist at the entry end of the market. Specialty Answering Service's cheapest tier is $44 a month before a single minute is used, at $1.54 a minute after. Ten minutes of calls costs $59.40. Dialzara gives you 60 minutes for $29, and ai-receptionist.com gives you 60 for $14. The human base fee alone exceeds the AI's entire monthly plan.
Choose the AI if: your after-hours calls are mostly the same six questions — are you open, do you cover my postcode, how much roughly, can I book Tuesday — and the job is to capture the enquiry rather than resolve it. That describes the large majority of after-hours volume for trades, clinics, salons, lettings and professional services.
When is an AI receptionist not worth it?
An AI receptionist stops paying for itself when your after-hours call volume is too low to clear the subscription, when your real problem is daytime calls rather than out-of-hours ones, when per-minute pricing collides with long calls, or when a wrong answer costs you more than voicemail would have. Four situations, in rough order of how often they come up.
Your volume is too low.
Eight distinct after-hours callers a month, a £120 average job, 40% margin and a 35% close rate gives you a monthly exposure of £134 — and that's the ceiling, assuming the AI answers every one and converts at your normal rate. A £149/month plan with a £199 setup fee loses you money on those numbers, comfortably. Run the arithmetic before you sign, and be honest about N.
You're already answering.
Businesses often discover, after running the export described above, that their real problem is calls missed between 9am and 5pm while someone is on a ladder or with a patient. That's a different problem with different fixes — call overflow, a shared inbox, a part-time coordinator — and an after-hours AI won't touch it.
Per-minute pricing and long calls.
Most AI vendors charge per minute. Talkdesk's 2024 benchmarking report put average contact-centre talk time at around 2 minutes 50 seconds (2023 data — contact centre, not SMB, and the sample size isn't disclosed). At that average, a 60-minute entry plan covers 21 calls. If your callers routinely run five or six minutes, it covers ten, and you're into overage by the second week. Per-call pricing — The VoIP Shop at £0.85, Smith.ai at $2.00 — is more predictable for long-call businesses.
A bad answer costs more than no answer.
This is the one vendors don't put in the ROI calculator. A voicemail is a neutral outcome. An AI that books the wrong slot, quotes a price you don't honour, or confidently tells a caller you don't cover their area when you do, produces a negative return — a lost job plus a complaint. Budget for the first fortnight of transcript review, and treat any vendor who tells you it works perfectly out of the box as someone who hasn't run one.
FAQ
How much does a missed call cost a small business?
It depends almost entirely on your average job value and close rate, which is why single-figure industry averages are misleading. Ambs Call Center's 2025 model puts it at $12.15 per missed call as a cross-industry average. For a business with a £250 average job, 50% margin and a 40% close rate, the figure is about £50; for a £2,000 average job at 70% margin and 30% close, about £420. Work it out from your own numbers rather than a benchmark.
Is it true that 62% of business calls go unanswered?
That figure comes from a 411 Locals study published on 18 January 2016, which monitored 85 businesses across 58 industries for 30 days and found 37.8% of calls answered live, 37.8% sent to voicemail and 24.3% getting no response. It's a real study with disclosed methodology, but it is a decade old, ran for a month, and was conducted by a marketing agency rather than a research firm. At least one vendor blog currently cites it as 2024 research. Treat it as historical context, not a 2026 benchmark.
Is an AI receptionist cheaper than a human answering service?
On published pricing, yes — about 5x cheaper per call on the only vendor that publishes both. Smith.ai prices its human tier at $10.00 a call and its AI tier at $2.00. We looked for a low-volume case where a human plan undercuts an AI subscription and couldn't find one: the cheapest 24/7 human option starts at a $44 monthly base fee before any minutes, which already exceeds several AI plans outright.
What happens to calls the AI can't handle?
Most vendors offer call transfer or escalation to a human, but check it's on the tier you're buying — Rosie's $49 Professional plan doesn't list live transfer, which first appears on its $149 Scale plan. Costs differ too: Bland AI charges $0.05 a minute for transfer time on top of its base rate. Before you buy, ask what triggers a transfer, where it routes at 2am, and whether transferred minutes get billed twice.
How long does an AI receptionist take to set up?
Most vendors in this category sell self-serve with a free trial rather than an implementation project, which tells you roughly how long setup takes. Trial lengths as published on 13 August 2026: ai-receptionist.com 3 days, Dialzara 7, Rosie 7, Frontdesk 7, Goodcall 14, RingCentral 14. Smith.ai offers a permanently free 25-call tier. The exception worth pricing in is Hand On Web, which charges a £199 setup fee on its £149 plan and a £499 setup on its top tier — a real implementation cost the others don't have. Budget your own time separately: the configuration is quick, but reviewing the first fortnight of call transcripts is where the quality actually comes from.
Will it book jobs into my calendar or CRM?
Some will, and this is the question that should drive your shortlist more than price. Integration depth varies enormously: Jobber's AI Receptionist is listed on its pricing page as a $29/month add-on to Jobber's own field-service platform and books straight into it, while Slang.ai is built specifically around restaurant reservation systems. (Jobber's product page and pricing page currently disagree on whether the add-on is included with its top tier — worth confirming before you budget for it.) Standalone vendors vary from "sends you an email summary" to full two-way calendar booking. Before you compare monthly prices, ask each vendor to demonstrate a booking landing in the specific system you already use — a receptionist that captures an enquiry you then have to re-key is worth considerably less than one that doesn't.
What happens to call recordings under UK GDPR?
Call recordings and transcripts are personal data, so running an AI receptionist means your vendor is processing personal data on your behalf and you need the usual apparatus: a lawful basis, a privacy notice that reflects what's actually happening, and a data processing agreement with the vendor. Three things worth asking before you sign: where the audio is processed and stored, how long recordings are retained by default, and whether your call data is used to train the vendor's models. That last one is not always opt-out by default. The ICO's guidance on call recording is the right starting point, and this is worth ten minutes with someone qualified rather than a blog post.
How long before an AI receptionist pays for itself?
For most service businesses, the first recovered job. At £99 a month against a £340 job at 55% margin, one converted after-hours enquiry returns £187 and covers roughly two months of subscription. Payback period is the wrong thing to worry about here. The question that actually decides it is whether your after-hours call volume is high enough to produce that one job reliably, every month, which is a question about N rather than about the vendor.
Do I need to tell callers they're speaking to an AI?
Probably, and the rules changed eleven days before this post was published. Article 50 of the EU AI Act became applicable on 2 August 2026. It requires that people interacting directly with an AI system are informed of that "from the start of the first interaction in a clear and distinguishable manner", unless it's obvious. The European Commission's guidance says that "obvious" exemption "should be interpreted in a restrictive manner" — a natural-sounding voice agent almost certainly doesn't qualify.
Read the allocation of duty carefully, because most write-ups get it backwards. Article 50(1) puts the obligation on the provider of the AI system — your vendor — not on the business deploying it. The deployer-facing duties in Article 50(3) and 50(4) cover emotion recognition and deepfakes, not voice-agent disclosure. So if you're a UK business running a vendor's AI receptionist, you are not the party the article names, whether or not your callers are in the EU. There's no equivalent blanket UK requirement either.
That's the legal position. The practical one is that it's your phone number, your brand and your greeting script, so ask your vendor what the greeting says by default and change it if it says nothing. Disclosure costs three words at the top of a call. None of this is legal advice and the area is moving quickly — check it with someone qualified before you rely on it.
Work out your own number first
Before you price a single vendor, export 90 days of call logs and count the distinct after-hours callers you didn't answer. That number decides everything downstream, and it takes twenty minutes to get.
Ten distinct after-hours callers a month at a £200 job and typical margins puts your exposure around £350 — comfortably ahead of a £99 plan even if the AI converts only a third of what you'd convert yourself. Below that, the arithmetic gets thin fast, and any vendor who tells you otherwise is selling.
- Oldroyd, McElheran & Elkington — "The Short Life of Online Sales Leads", HBR 89(3), Mar 2011
- Ambs Call Center — "The Real Cost of a Missed Call", 8 Sep 2025 · press release, 21 Aug 2025
- 411 Locals — "Small Business Owners Don't Answer 62% of Phone Calls", 18 Jan 2016
- Talkdesk — 2024 Global Contact Center KPI Benchmarking Report
- European Commission — Transparency obligations under Article 50 of the AI Act
- Vendor pricing pages, all retrieved 13 August 2026 — linked inline in the pricing tables above
Used as claims, not as evidence: the "$126,000 annual loss" figure (untraceable — one attribution 404s, another names a different company) and the "62% of calls unanswered" figure (real, but a 30-day study of 85 businesses from January 2016). Both are examined in the post; neither underpins any recommendation in it.