Loading
Most speed-to-lead advice measures the wrong clock. The classic studies tracked outbound callbacks on web forms, not the phone ringing on your desk. Here is what the research actually proves for service businesses, what callers do when nobody picks up, and which widely quoted statistics do not survive a source check.
Last updated: July 2026. In 2024, researchers posing as prospective clients contacted 500 US law firms. Only 40% of those firms answered the phone, down from 56% in 2019 (Clio, Legal Trends Report 2024). Those firms were not short of leads. They were short of pickups. This guide separates what the speed-to-lead research actually proves from what gets repeated online, and translates it for a business whose leads arrive as a ringing phone.
Key takeaways
- Speed to lead is the time between a prospect raising their hand and a human (or a capable AI) responding. For service businesses, most of that clock runs while the phone is ringing.
- The famous 5-minute rule comes from 2007 research into outbound callbacks on web-form leads. It measured the odds of reaching and qualifying a lead, never closing one.
- In 2026, call-tracking platform Invoca analyzed more than 70 million calls and found only 56% of callers reach a person. That rises to 71% on calls lasting over 30 seconds.
- In 2026, 75% of consumers said they had hung up after being left on hold too long, and 79% said they would switch to a competitor that responds faster (Invoca, B2C Buyer Experience Report 2026, n=1,356).
- Several of the most-quoted speed-to-lead statistics do not survive a source check. We name them below, with a sourced replacement for each, and we disclose our own sourcing conflicts too.
Speed to lead is the elapsed time between a prospect making contact and your business responding with a real, useful reply. It is a response-time metric, not a marketing metric. In 2026, Invoca's analysis of more than 70 million inbound calls found that only 56% of callers to businesses actually speak with a person (Invoca, Lead Conversion Benchmarks Report 2026).
Most articles on this topic were written for inside-sales teams. They assume a lead is a form submission sitting in a CRM, waiting for a rep to dial back. That describes a software company. It does not describe a plumbing company.
For a service business, the lead is usually a person calling you right now, with a burst pipe or a cracked tooth or a lease question. Your speed to lead is decided in the few seconds before someone picks up. Get that wrong and there is no callback window to optimize, because the caller already dialed the next name on the list.
Service businesses run two separate response clocks, and almost all of the published research covers only one of them. Confusing the two is the most common mistake in speed-to-lead advice.
The callback clock starts when someone fills in a form, requests a quote, or submits a Google or Meta lead ad. It stops when you call them back. This is the clock that Harvard Business Review and the MIT-affiliated lead response research both measured.
The pickup clock starts when the phone rings. It stops when a person answers. No research firm has ever needed to prove that a 30-second pickup beats a 3-minute one, because the caller simply hangs up and calls a competitor.
The pickup clock is the one that matters most for the trades, clinics, and firms we work with, and it is the one nearly every speed-to-lead article ignores. Close to 70% of Americans tend to use phone support for customer service, even though only 35% say they prefer it (YouGov Profiles, March 2025). Preference and behavior are not the same thing. People still call.
The 5-minute rule comes from a single 2007 study, and it proves less than most people claim. Dr. James Oldroyd, then a faculty fellow at MIT Sloan, analyzed three years of data covering more than 15,000 leads and 100,000 call attempts across six companies. The finding: the odds of contacting a lead drop by 100× when you call at 30 minutes instead of 5 minutes, and the odds of qualifying one drop by 21× (Lead Response Management Report, presented at MarketingSherpa's B2B Demand Generation Summit, 2007).
Read those verbs again. Contact and qualify. That study states plainly that it did not address close ratios. Every blog post promising that the 5-minute rule lifts your conversion rate by some precise percentage has quietly swapped in a verb the research never used.
The 2007 report carries a less dramatic figure that we find more useful: within the first hour, the odds of contacting a lead fall by more than 10× and the odds of qualifying one by more than 6×. That is still an enormous decay curve, and it survives scrutiny. The survey half of the same report, covering 495 companies, found that 38% could not say how long they took to make a first call attempt, because they did not know, did not measure it, or did not answer the question.
Slower than almost any owner would guess. Harvard Business Review researchers audited 2,241 US companies by submitting a web inquiry to each one. Nearly a quarter, 23%, never responded at all. Among the companies that did reply within 30 days, the average response time was 42 hours (Harvard Business Review, "The Short Life of Online Sales Leads," 2011).
The distribution of those HBR response times is worth knowing: 37% responded within an hour, 16% took between one and 24 hours, and 24% took longer than a day.
"Are you confident that your company is effectively handling potential customers' online queries? Think hard. Our research shows that most companies are not responding nearly fast enough."
James B. Oldroyd, Kristina McElheran and David Elkington, Harvard Business Review, March 2011
A second dataset in the same HBR article, covering 1.25 million leads at 29 B2C and 13 B2B companies, found that firms trying to make contact within an hour were nearly 7× as likely to qualify the lead as those that tried an hour later, and more than 60× as likely as those who waited 24 hours or more. That is the real home of the 60× figure you see quoted around the web.
The HBR audit and the 2007 lead-response study are both strong. Both are about web-form leads, and both are more than a decade old. Treat them as the physics of lead decay, not as a benchmark for your phone line.
Roughly half. Invoca's 2026 benchmark, built on more than 70 million calls and 600 million conversation minutes across ten industries, found that 56% of callers speak with a person. Answer rates climb to 65% on calls lasting more than 15 seconds and 71% on calls over 30 seconds (Invoca, Lead Conversion Benchmarks Report 2026).
The 15-second qualifier in that Invoca data matters, and most vendors leave it out. A slice of unanswered calls are misdials and instant hang-ups, not lost customers. Work back from the 71% figure and roughly 29% of calls that last long enough to be real go unanswered. The honest read is somewhere between a quarter and a third of genuine inbound calls, not two-thirds.
The most-quoted figure in this field says otherwise: "62% of small business calls go unanswered." It traces to a single 2016 study by a local SEO vendor that monitored the phones of 85 small businesses for 30 days. It is real, and it is worth knowing, but it is 85 businesses and a decade old. We have seen it republished on dozens of sites as though a dozen studies agree. They do not. There is one study.
They leave, and they rarely leave a message first. In 2026, 75% of consumers said they had hung up after being put on hold too long, and 79% said they would switch to a competitor that responds faster (Invoca, B2C Buyer Experience Report 2026, n=1,356 US and UK consumers, fielded May 2026). The hold-up figure was reported as a 26-point rise year over year, which is a very large single-year move in survey data, so read it as a direction of travel rather than a precise measurement.
Voicemail is not the safety net owners assume. Invoca's home-services data shows that fewer than 3% of callers pushed to voicemail leave a message (Invoca, 2024). More than 97 out of 100 leave nothing behind.
Unanswered calls are where the money leaks, and the leak is invisible on every report you look at. A missed call leaves no lead record, no form entry, and no CRM row. You cannot see the customer you never met. For the arithmetic on what that costs by industry, see our breakdown of what missed calls really cost a business.
It varies more by trade than by company size. The strongest industry-level evidence comes from secret-shopper studies, where researchers pose as customers rather than asking businesses to self-report.
| Industry | What the research found | Source |
|---|---|---|
| Law firms | Only 40% of 500 firms answered the phone, down from 56% in 2019. Counting firms that also failed a follow-up attempt, 48% were essentially unreachable by phone. Of firms that missed a call, just 20% ever returned it. | Clio, Legal Trends Report 2024 (secret shopper study), as reported by Clio and the Oklahoma Bar Association |
| Real estate | 47% of online property inquiries at 25+ brokerages were ignored entirely. Average response time 8 hours 17 minutes; median 39 minutes. | Mike DelPrete, Scholar-in-Residence, University of Colorado Boulder, 2024 |
| Home services | Answer rates ranged from 32% to 74% depending on the trade, across nine sub-industries including HVAC, plumbing and restoration. | Invoca, Home Services Benchmarks 2026 |
| Dental | About 62% of dentists named staffing shortages as the biggest challenge facing their practice in 2025. | ADA Health Policy Institute, surveyed late 2024, published February 2025 |
| All industries | 56% of callers speak with a person; 71% on calls over 30 seconds. | Invoca, Lead Conversion Benchmarks 2026 |
Two details in the industry table deserve attention. First, the real estate mean and median are wildly apart: 8 hours versus 39 minutes. That gap says most agents reply reasonably fast and a long tail never replies at all, which is a different problem from being uniformly slow. Second, the legal answer rate moved the wrong way over five years, from 56% down to 40%, during a period when every firm had more technology available, not less.
The dental row is the odd one out, and deliberately so. The ADA measured how dentists rank their own challenges, not how many calls they miss. We include it because it points at the cause rather than the symptom: a front desk that cannot hire is a front desk that cannot answer. Treat it as context, not as a call statistic.
If you run one of these businesses, the vertical breakdowns go deeper: AI reception for law firms, for real estate, for dental practices, and for HVAC and plumbing.
Several of the most-quoted numbers in this field do not survive a source check. We went looking for the original studies behind the figures that dominate this topic's search results. Most dead-end in vendor blogs citing other vendor blogs. One turned out to be real research whose publisher no longer exists. Either way, you cannot build a business case on a number you cannot read.
| Commonly quoted claim | Problem | Use this instead |
|---|---|---|
| "391% more conversions if you call within a minute" | This one does have an origin, which is why the blanket "it is made up" dismissal is wrong. It traces to a Velocify whitepaper, "The Ultimate Contact Strategy," from around 2013. The catch: Velocify was acquired and folded into Ellie Mae, and the original report is offline, so nobody quoting it today can read it. We could not read it either. Secondary accounts consistently describe it as call-center platform data rather than independent research, and the comparator drifts between retellings. | The 2007 contact and qualify odds, stated with the correct verbs. |
| "85% of callers never call back" | Variously credited to Forbes or BIA/Kelsey with no report title, year, or sample size anywhere. | 79% would switch to a competitor that responds faster (Invoca, 2026). |
| "78% of customers buy from whoever responds first" | Attributed to a "Lead Connect survey" that has no published methodology. Also circulates falsely badged as National Association of Realtors data. | The HBR 7× and 60× qualification figures, which are real and measured. |
| "43% of hot leads vanish within 24 hours" | No primary source. Appears to be a garbled retelling of HBR's 60× finding. | HBR, 2011: 23% of companies never respond at all. |
| "$100 to $1,200 lost per missed call" | Vendor arithmetic presented as research. The real figure depends entirely on your average job value. | Run your own numbers against your average ticket. |
Two disclosures, because a post that audits everyone else's sourcing should publish its own. Four of the twelve live sources listed at the end of this article come from Invoca, a call-tracking company whose commercial interest is the same missed-call story we are telling. We used their data because the sample sizes are large and the methodology is disclosed, and because the direction matches what the independent secret-shopper studies from Clio and Mike DelPrete found. Read those four figures with that in mind.
The second disclosure is about us. We sell AI receptionists. We have the same incentive as every vendor quoting the 391% figure, and being aware of a bias is not the same as being free of it. What we are offering is not neutrality, which we do not have. It is that every number here names its publisher, its year, and where possible its sample size, so you can check our work rather than trust us.
You fix it by removing the human bottleneck from the first response, not by asking your team to try harder. Here is an honest comparison of the realistic options, including where each one falls down.
| Option | Typical time to first response | Nights and weekends | Books the appointment | Honest limitation |
|---|---|---|---|---|
| Voicemail | Hours to never | Records only | No | Under 3% of callers leave a message |
| Call them back yourself | Minutes to days | No | Yes | Depends on you being free, which during a job you are not |
| In-house receptionist | Seconds, when free | No | Yes | One call at a time, 40 hours a week, plus sick days and turnover |
| Human answering service | Seconds to minutes | Usually yes | Sometimes | Often takes a message rather than booking; per-minute billing |
| Missed-call text-back | Seconds, after the miss | Yes | No | Fires only once the call is already lost; poor fit for emergencies |
| AI receptionist | Under 3 seconds | Yes | Yes | Hands off sensitive or unusual calls to a person |
Missed-call text-back deserves a note, because it is often sold as the cheap fix. It is genuinely useful, and it is a downgrade. It converts a lost call into a text thread, which works for a quote request and fails badly for a burst pipe at 11pm. It is a recovery tool, not a speed-to-lead tool.
The callback clock still needs an owner. Answering the phone does nothing for the form fills, Google Local Service leads, and Meta ad inquiries sitting in your inbox, and that is exactly where the 5-minute research applies. Whoever or whatever handles those needs a defined target measured in minutes, not "when someone gets a chance." Outbound follow-up is available on our Growth and Scale plans for that reason.
Here is the arithmetic we promised, using your own numbers rather than a borrowed statistic. Say you take 200 calls a month and miss 25% of them. That is 50 missed calls. If 1 in 4 of those callers would have booked, and your average job is $500, the exposure is 12 or 13 jobs and about $6,250 a month. Halve the miss rate and halve the booking rate, so 25 missed calls and 1 in 8 booking, and it is still a little over $1,500. Run it with your real call volume and your real ticket, because that is the only version that matters. Our AI receptionist cost guide breaks the pricing side down.
If you are weighing the staffed options against each other, we compared them directly in AI receptionist versus answering service.
When the conversation is emotional, high-stakes, or genuinely unusual. This is where speed stops being the goal. In 2026, 83% of consumers said they value human connection during high-stakes purchases, and 59% said they prefer a human when both options are available (Invoca, B2C Buyer Experience Report 2026).
A bereaved family calling a law firm, a patient describing a frightening symptom, a tenant who has already complained twice: these calls need a person, and answering them in three seconds with the wrong voice is worse than answering them in thirty with the right one.
The practical answer is not to choose. Answer everything instantly, then route the calls that need a human to a human, with the context already captured. In the same 2026 research, 63% of US consumers said they could not reliably distinguish AI from human interactions, and 83% said AI should clearly identify itself. Both things can be true: disclose it, and make the handoff clean. We covered the acceptance question in detail in will customers accept an AI receptionist.
We built AIEmply around the pickup clock, because that is where service businesses lose. The AI answers in under 3 seconds, compared with the 2 to 3 minutes typical of manual handling, and it answers 100% of calls against roughly 60% managed manually. On that 60% figure, be skeptical of us the same way we asked you to be skeptical of everyone else. Invoca's 70-million-call dataset puts the raw connect rate at 56%, and the generous read earlier in this article put genuine answered calls nearer 71%. Our 60% sits between those two, which is where we think an honest manual baseline lands. Nights, weekends and holidays are covered, and unlimited simultaneous calls mean a Monday morning rush never produces a busy signal.
Answering is the floor, not the product. During the call the AI qualifies the lead, books into your live calendar, updates your CRM, and sends a confirmation text. Known spam and robocalls get filtered out, so your team is not interrupted by extended-warranty calls. Anything that needs a person goes to one, by warm handoff with a full summary attached, so the caller never repeats themselves. Coverage runs to 50+ languages with mid-conversation switching.
Setup is a 15-minute consultation, then 4 to 7 days of configuration, then 3 to 5 days of testing. Test-ready in 1 to 2 weeks, handled by our team.
One part of the setup process is worth calling out because most owners underestimate it. A real person on our side reads call transcripts and refines the script every week. An AI receptionist that nobody tunes plateaus fast. Call 10,000 should be better than call 1, and it only gets there if someone is actually reading the transcripts.
Plans are flat monthly: Starter at $149/mo, Growth at $399/mo, and Scale at $599/mo, with 30% off annual billing. Full detail is on the pricing page. The guarantee is simple: "Billing starts only after your AI Employee is live. If the first month delivers no measurable result, the next month is free." No credit card is required to start. AIEmply is built by Veltro Systems LLC.
Speed to lead is real, and most of what is written about it is measuring the wrong clock for your business. The research proves that the odds of reaching and qualifying a lead collapse within the first hour. It does not prove a tidy conversion percentage, and anyone quoting one is repeating a number nobody can source.
For a service business, the useful version is simpler. Roughly half of callers to businesses reach a person. Fewer than 3% of the rest leave a voicemail. Nearly 8 in 10 will try whoever answers next. You do not need a study to act on that. You need the phone answered.
100% Answer Rate • Ready in 1–2 Weeks • Performance Guarantee
If you want to see what your own pickup clock looks like, try a live demo, compare plans on the pricing page, or book a 15-minute consultation and we will walk through your call volume with you. Billing starts only once your AI Employee is live.
Speed to lead is the time between a prospect contacting your business and you responding. For businesses that get leads by phone, most of that time is spent with the phone ringing. In 2026, Invoca found only 56% of callers to businesses reach a person, rising to 71% on calls lasting over 30 seconds.
Partly. The 2007 Lead Response Management study found the odds of contacting a lead drop 100× and the odds of qualifying one drop 21× when you call at 30 minutes instead of 5. That study explicitly did not measure close rates, so any claim that the rule lifts conversions by a set percentage goes beyond the evidence.
Around 44% of callers do not reach a person, based on Invoca's 2026 analysis of more than 70 million calls. That figure includes misdials and instant hang-ups, so the true rate for genuine inbound calls is lower. The widely quoted 62% comes from one 2016 study of just 85 small businesses.
There is no credible source for it. The claim is attributed variously to Forbes or BIA/Kelsey with no report title, year, or sample size. A sourced alternative from 2026: 79% of consumers say they would switch to a competitor that responds faster, and fewer than 3% of callers sent to voicemail leave a message.
Before the caller gives up, which in practice means within a few rings. There is no benchmark study on ideal pickup time because the behavior is obvious: 75% of consumers said in 2026 they had hung up after being left on hold too long. An AI receptionist answers in under 3 seconds, against 2 to 3 minutes typical of manual handling.
It helps, but it works after the call is already lost. Text-back turns a missed call into a text thread, which suits a quote request and fails for an emergency at 11pm. It is a recovery tool rather than a response-time fix. Answering the call in the first place removes the need for it.
Under 3 seconds, and yes. The booking part is the distinction worth checking when you compare providers, because most answering services take a message rather than completing the appointment. AIEmply closes the booking during the call, writes it into your live calendar, and confirms by text. Plans start at $149 per month.