We Fact-Checked the 5 Statistics Every AI-Receptionist Vendor Quotes

How did we actually check these five statistics?

If you've researched AI receptionists, missed-call cost calculators, or "why isn't my phone getting answered" content for more than a few minutes, you've seen these five numbers. They appear, often word for word, across vendor blogs, LinkedIn posts, and sales decks, usually with no link to an actual study. We traced each claim back to its original source, then independently re-checked that tracing against the earliest document we could find, rather than trusting the last blog that repeated it. Here's what actually turned up.

Is it true that 62% of calls to small businesses go unanswered?

This is the one claim on this list that traces to an actual study, so we'll say that upfront: it holds up better than the other four, with real caveats. The number comes from 411 Locals, a local-SEO marketing agency, in a self-published company-blog post dated 18 January 2016. They monitored 85 businesses across 58 industries for 30 days: 37.8% of calls were answered live, 37.8% went to voicemail, and 24.3% got no response of any kind. Add the voicemail and no-response figures together, 37.8 plus 24.3, and you get 62.1%, the arithmetic root of the "62%" figure that's been repeated ever since.

The honest way to cite this number is as narrow as the study that produced it: a 2016 vendor study of 85 businesses found 62% of calls did not reach a live person. It is not an ongoing industry benchmark. It was never peer-reviewed or independently audited, the sample is small and self-selected, and it was published by a marketing agency writing about the exact problem it sells a solution for. All of that can be true and the underlying finding can still be directionally real. What it cannot be is a citation for "the current unanswered-call rate for small businesses," because it was never designed to measure that on an ongoing basis.

Do 85% of callers who don't reach you really never call back?

No traceable source exists for this one. We checked the usual candidate origins for this figure, including PATLive and other telecom-industry material sometimes cited for it, and the 411 Locals 2016 study itself, and found no study, survey, or dataset anywhere in the chain. The number circulates purely by mutual restatement: one vendor blog cites "research shows," the next blog cites the first blog, and the specific figure hardens into fact through repetition rather than evidence.

There's a reasonable underlying intuition: callers who hit a dead end probably do often go elsewhere. But "85%" specifically has no findable origin. The honest framing is to name it for what it is, a number that's everywhere and traces to nowhere, not a statistic to repeat as fact.

Did missed calls really cost businesses $75 billion a year?

This is the most consequential misattribution on the list, because the underlying study is real, it just isn't measuring what it's cited for. The actual source is NewVoiceMedia (now part of Vonage), a report called "Serial Switchers," published May 2018, based on an Opinion Matters survey of 2,002 US internet users aged 16 and over, fielded January to February 2018. That study measures the cost of customers switching brands after poor customer service broadly: rudeness, being transferred repeatedly, long hold times, feeling unappreciated. Phone calls are one input among several, and the study does not isolate missed or unanswered calls as the specific loss driver.

It gets stranger. NewVoiceMedia's own landing page for this research still shows the predecessor 2016 figure, $62 billion, not $75 billion. The $75 billion number circulating in AI-receptionist marketing appears to be a later escalation of a figure that was never about missed calls specifically in the first place. The honest framing: this is a real, named 2018 consumer-survey statistic about broad customer-service-driven brand switching, not a study of missed phone calls.

Do 80% of callers sent to voicemail really not leave a message?

This one has a citation trail we could actually follow, and it dead-ends four hops back with no original data anywhere in it. Modern vendor blogs attribute it to "Marchex" or "BIA/Kelsey" research, but Marchex's own published reports contain no such figure. Following the trail further back leads to a November 2014 CRM Magazine article that cites an unlinked "Forbes" source, which leads to a July 2014 Forbes contributor piece that in turn attributes the number to "The New York Times, 2009," again with no link. The 2009 New York Times piece it eventually points to is, by every secondary account of its contents we could find, an anecdote-driven lifestyle piece, not a data-driven study.

Four hops, zero original datasets. This is a textbook case of a statistic that sounds precise, "80%" reads like it came from somewhere specific, but the citation trail simply stops in a 15-year-old lifestyle article that was never measuring this in the first place.

Does a missed call really cost a business $1,200 on average?

Almost every vendor page that uses this figure attributes it to "Invoca research." We checked three current, live Invoca reports directly. None of them contains a $1,200 figure. This has the signature of vendor arithmetic (an assumed average job value multiplied by an assumed close rate) that got laundered into "Invoca research says" by secondary blogs with no locatable citation back to an actual Invoca publication.

Do not cite this as an Invoca finding under any framing. If you want to build your own missed-call cost estimate, build it transparently from your own average deal value and your own close rate, and say that's what you did, rather than borrowing an unsourced average that happens to sound authoritative.

So which of these five statistics actually holds up?

One traces to a real, if narrow and dated, study. One is a genuine study measuring something else entirely. Three have no locatable source at all. Here's the summary:

Why do unverified statistics keep circulating like this?

Every one of these five numbers followed the same pattern. A vendor blog states a statistic with confident phrasing, cites another vendor blog or a vaguely named source, and the next blog down the chain repeats it with the same confidence and one fewer link back to anything real. None of the five started as an outright fabrication. Four of them either started as a real study measuring something narrower than what it's now cited for, or started nowhere findable at all and simply accumulated authority through repetition. The pattern isn't unique to AI receptionists, it shows up anywhere a scary statistic helps sell a product, but this niche has more of it than most because "how many calls are you missing right now" is precisely the fear a receptionist vendor is selling against.

What do we publish instead of numbers like these?

Our answer to "how many calls actually get answered" isn't a 2016 study about other businesses, it's our own production data. We ran 1,097 real AI outbound calls in the UAE between June and July 2026 and logged every one automatically: 47.9% connected for 10 seconds or longer, 21.1% became genuine 30-second-plus conversations, and just 0.16%, one call out of 634 scored for sentiment, came back hostile. Those numbers aren't a comparison to anyone else's marketing claim. They're what actually happened on our own lines, with the methodology stated openly enough that you can judge the sample size for yourself.

If you're evaluating whether an AI receptionist deployment is worth it for your business, ask any vendor quoting you a statistic the same question we asked ourselves here: where does that number actually come from, and what was it originally measuring? For a longer first-hand account of what running AI on every call actually looks like day to day, see our month of letting AI answer every call. Our services page lays out what we build and how we measure it once it's live.