Skip to content
All posts

22 min read

Speed to Lead in 2026: The Real Benchmarks, and Why an Instant Reply Isn't a Response

Most lead response time statistics are recycled from 2011 or invented. Here are the real benchmarks, dated and sourced — and the metric that beats speed.

Content DeskA dedicated editorial team responsible for creating, reviewing, and publishing insightful content across technology, business, and product updates. We focus on clarity, accuracy, and practical value for readers.
Lead response decay: under five minutes is 21x more likely to qualify, 30 minutes is baseline, after 24 hours the lead is cold.
The 21x figure comes from a 2007 study of more than 15,000 leads. It is still the best data we have — and it never measured close rates, whatever you may have read elsewhere.

Either you answer in two minutes, or you never answer at all

In March 2024, RevenueHero ran an experiment that deserves to be better known. They submitted a demo request — a genuine hand-raise, the highest-intent action a visitor can take — to 1,000 B2B SaaS companies, and then they waited.

635 of them never replied. Not slowly. Never.

Of the 365 that did reply, the average response time was one day, five hours and seventeen minutes. But that average hides the real finding, which is stranger and more useful: 172 of those 365 replied in under two minutes. Another 31 came within the hour. The rest trailed off into days and weeks.

There is no middle any more. There is a tier of companies that answers before the visitor has closed the tab, and a much larger tier that never answers at all. Almost nobody occupies the ground in between — and if you are in that larger tier, you are not competing on price, product or positioning. You are not competing.

This post is about what the evidence actually says, where the widely-quoted numbers come from, and why the metric nearly everyone optimises is the wrong one.

First, a warning about the numbers you will read elsewhere

If you search "lead response time statistics 2026", you will find dozens of pages presenting confident, precise, freshly-dated figures. We read a great many of them while researching this post. Most are recycled, misattributed, or invented.

Some specifics, because vague warnings are useless:

  • "The average B2B lead response time is 42 hours." This is real, and it is from the Harvard Business Review, in March 2011. It is fifteen years old. It is not a 2026 benchmark.
  • "Real estate agents take 917 minutes to respond." Real, and from the WAV Group Agent Responsiveness Study published in January 2014.
  • "62% of calls to small businesses go unanswered." Usually credited to BIA/Kelsey in 2023 or 2024. It traces back to a 30-day study of 85 businesses, run by 411 Locals in 2016. Credit to OnCrew, a vendor that does its own source-tracing and flags this one publicly.
  • "Missed calls cost the average small business $126,000 a year." No study exists. It is arithmetic: an assumed number of missed calls multiplied by an assumed value per call.
  • A cluster of pages cite a "Forrester B2B Sales Automation Index, Q1 2026" for figures on AI SDR adoption and cost per qualified opportunity. We could not find any such index. Several of those same pages attribute figures to The Bridge Group's SDR report that do not appear in the report it actually publishes.

We point this out for a self-interested reason as much as an honest one. If you are going to change how your business handles enquiries, you deserve to know which numbers are load-bearing. Everything below is dated, attributed, and linked to the original.

What the research actually says

The 5-minute rule, and what it really measured

The most-cited statistic in sales is the "5-minute rule", and it comes from work by Dr. James Oldroyd, presented at MarketingSherpa's B2B Demand Generation Summit in October 2007 and published through InsideSales.com. The methodology is stated plainly in the study itself: three years of data across six companies, more than fifteen thousand leads and over one hundred thousand call attempts.

Its headline findings:

Delay

Effect

5 min → 30 min

Odds of contacting a lead drop 100x

5 min → 30 min

Odds of qualifying a lead drop 21x

5 min → 10 min

Odds of contacting drop 5x; dial-to-qualify drops 4x

Within the first hour

Contact odds fall 10x; qualification odds fall 6x

Beyond 20 hours

Additional dials actively reduce your odds of contact and qualification

Two rows in that table matter more than the famous 100x.

The first is the 5-to-10 minute row, which almost nobody quotes. The decay is not a cliff at five minutes with a flat plain before it. You lose 4x on qualification odds between minute five and minute ten. Whatever your target response time is, the curve is steepest at the very beginning.

The second is the 20-hour finding. Past that point, chasing harder makes things worse — every additional dial reduced the odds of making contact. Persistence is not a substitute for speed, and after a certain point it is the opposite of one.

And one thing the study did not do, in its own words: "This study did not address close ratios." It measured contact and qualification odds. Any page citing "MIT" for close rates or revenue impact has invented that connection. We are not going to repeat it here.

The audit that showed how few companies respond at all

In March 2011, Oldroyd, Kristina McElheran of Harvard Business School, and David Elkington published The Short Life of Online Sales Leads in Harvard Business Review. They audited 2,241 US companies by submitting a web-generated test lead to each.

  • 37% responded within an hour
  • 16% responded within one to 24 hours
  • 24% took longer than 24 hours
  • 23% never responded at all
  • Among companies that did respond within 30 days, the average was 42 hours

A second dataset in the same work covered 1.25 million leads across 29 B2C and 13 B2B companies. Firms making contact within an hour were roughly 7x more likely to qualify the lead than those contacting an hour later — and more than 60x more likely than those waiting 24 hours or more.

Fifteen years later, it is worse

Set the 2011 audit beside the 2024 one. In 2011, 23% of companies never responded. In RevenueHero's 2024 test of 1,000 B2B SaaS companies, 63.5% never responded — and that test counted automated replies as responses.

Only 113 of the 1,000 companies had a scheduling tool on the page at all.

The gap between the top tier and everyone else has widened enormously. The tools to respond instantly have never been cheaper or more available, and the majority of the market has gone backwards.

Benchmarks by industry, honestly dated

Industry

Finding

Source

Date

B2B SaaS

63.5% never respond; 47% of responders reply in under 2 min

RevenueHero, n=1,000

Mar 2024

Cross-industry

23% never respond; 42h average among responders

HBR, n=2,241

Mar 2011

Real estate

48% of enquiries never answered; 917 min average

WAV Group, n=384 brokers

Jan 2014

Automotive

Industry average Internet Lead Effectiveness score of 33/100; dealers responded quickly by email or text 27% of the time

Pied Piper PSI, n=3,957 dealerships

2024–25

The real estate and automotive numbers are old, and we are labelling them as such. They are the best genuine mystery-shop data that exists for those industries. Pied Piper is the exception worth watching, because it still runs annually against nearly four thousand dealerships — in its 2025 edition, 19% of dealerships scored under 40, meaning no personal response at all.

The metric everyone optimises is the wrong one

Here is the problem with every study above, including the ones we are leaning on.

They measure time to first response. RevenueHero's methodology says so explicitly: automated replies counted. And an automated reply is trivially fast. "Thanks for your enquiry, someone will be in touch shortly" arrives in zero seconds and does nothing at all. If your metric is time-to-first-response, an autoresponder gives you a perfect score and zero additional revenue.

The number that matters is time to qualified response: how long before the visitor gets a reply that actually moves their decision forward.

A reply counts as a qualified response if it does four things:

  1. Answers the question they actually asked — from your real pricing, availability, terms or stock, not a generic acknowledgement.
  2. Captures something you did not have before — a name, an email, a budget, a timeline, an address, a use case. Ideally captured during the conversation, as it surfaces, rather than gated behind a form at the end.
  3. Says what happens next, specifically. "A member of our team will call you before 10am tomorrow" is a qualified response. "We'll be in touch" is not.
  4. Reaches a human when it needs to — quickly, with the context already attached, so nobody re-asks what was already answered.

Measured this way, most businesses that believe they respond in seconds are actually responding in hours or days. The autoresponder fires instantly; the qualified response waits for someone to open the CRM on Monday.

This is the number worth putting an SLA against, and it is the one we built ConGreeto to compress.

The after-hours problem, and an honest note about its statistics

You will have read that some large share of enquiries — 40%, 52%, 65%, take your pick — arrives outside business hours.

We went looking for the study behind those figures and could not find one. Every instance traced back to another vendor blog. OnCrew's source-tracing reaches the same conclusion about the closely-related "42–48% of home-service calls are after-hours" claim: it appears only in vendor content, with no primary study behind it.

So we are not going to give you a percentage. The behavioural data is better anyway, and it is properly sourced.

From CallRail's 2025 consumer survey of 1,000 US consumers:

  • 78% have abandoned a business after a call went unanswered
  • 82% will call a competitor instead
  • Only 42% leave a voicemail
  • 41% hang up after one to two minutes on hold
  • 24% switch to online chat instead
  • 52% consider an after-hours AI response better than no response at all

And from Jobber's 2026 Home Service Trends Report: 41% of jobs booked online arrive after hours, and 56% of homeowners expect a response within one hour.

Read those together and the after-hours argument stops being about volume and becomes about defection. The question is not what share of your enquiries arrive at 11pm. It is what those people do when nobody answers — and the answer is that four in five call a competitor, and fewer than half leave you any way to call them back.

The lead does not wait until morning. It goes somewhere else, and you never find out it existed.

Why this matters more in 2026 than it did in 2024

Something has changed about the traffic reaching your site, and it raises the cost of fumbling any single visitor.

Pew Research Center tracked the real browsing behaviour of 900 US adults across 68,879 Google queries, 12,593 of which returned an AI summary. Users clicked a traditional search result on 8% of visits where an AI summary appeared, against 15% where it did not — roughly half. They clicked a link inside the AI summary on just 1% of visits. And they ended their browsing session entirely on 26% of AI-summary pages, against 16% without.

Fewer people are arriving. But the ones who do arrive are different.

Ahrefs published its own data in June 2025: AI search accounted for 0.5% of their traffic but 12.1% of their signups — a conversion rate roughly 23x that of traditional organic. Those visitors viewed 50% more pages per session. Their own analyst, Patrick Stox, is candid about the flip side: users from AI search click links 75% less often, and if AI search becomes the default, total search traffic could fall to a fraction of today's.

HubSpot's 2026 State of Marketing report, surveying more than 1,500 marketers, frames the consequence well: buyers now "come later in the process and will be better-educated, with higher intent." In the same survey, 40% named lead quality or MQLs their single most important success metric — more than any other metric — and 93.8% said lead quality had improved over the past year.

The arithmetic is not complicated. You are getting fewer visitors. Each one is further along, better informed, and worth more. The cost of the one you fail to answer has gone up, and it will keep going up.

When fast AI response makes things worse

We sell an AI product. We are still going to tell you where this goes wrong, because you will find out either way and it is cheaper to find out here.

People say they do not want this. SurveyMonkey surveyed 2,017 US adults in December 2025: 79% strongly prefer human interaction over AI agents, 56% have negative feelings about companies using AI in customer experience, 50% would cancel a service that was solely AI-driven, and 89% believe companies should always offer a human contact option. Even among Gen Z, only 14% prefer AI at equal speed and quality.

Telling people it is a bot destroys most of the benefit. A randomised field experiment on more than 6,200 customers, published in Marketing Science in 2019 (Luo, Tong, Fang & Qu), found undisclosed chatbots were as effective as proficient human agents and four times more effective than inexperienced ones — and that disclosing the bot's identity up front cut purchase rates by 79.7%.

We think the ethical resolution to that finding is also the commercially correct one, and it is the entire basis of how ConGreeto is positioned: do not build a bot that impersonates your salesperson. Build an assistant that is obviously an assistant, that is genuinely useful in its own right, and that hands off to a human quickly and cleanly. The trust cost of disclosure is real. The trust cost of being caught is total.

The handoff is where deployments die. In the same SurveyMonkey research, only 15% of consumers report ever experiencing a seamless AI-to-human handoff. That is the actual product gap in this category, and closing it is worth more than another point of answer accuracy. One commenter on a Hacker News thread about customer service bots put the failure mode precisely: "They ask for info, then when I get a human, they ask for the same information again."

Vendors are overpromising, and buyers know it. From a discussion of Klarna's AI reversal: "Rough order of magnitude you can expect to deflect around 30% of inbound contacts with an AI chatbot... vendors are telling companies they can eliminate 80-90% of their customer service agent jobs with AI, and that is nonsense." Klarna itself is the cautionary tale — its AI was announced in February 2024 as doing the work of 700 agents, handling 2.3 million chats in the first month, and the company later began rehiring humans. Notably, the bot still handles two-thirds of enquiries. It was not a technology failure. As CEO Sebastian Siemiatkowski put it: "cost unfortunately seems to have been a too predominant evaluation factor... what you end up having is lower quality." His conclusion: "I just think it's so critical that you are clear to your customer that there will be always a human if you want."

And a badly-configured bot is expensive in ways nobody warns you about. One developer described a deployment where consultants had the assistant fire a greeting on every page load: "The next month they were surprised they got a $2000 bill for the API use and at first wondered if the bot was really popular." Another designer wrote about a bot that "confidently provided wrong opening hours for months."

Every one of those is avoidable. None of them is avoided by default.

What good actually looks like

The most credible published example of AI-assisted lead response we found comes from Salesforce's own sales organisation, in the 7th edition of its State of Sales report (4,050 sales professionals across 22 countries, fielded August–September 2025). Adam Alfano, who leads Salesforce's own sales organisation, describes putting agents onto leads nobody was working:

"We used to let these leads fall to the floor like sawdust. Now, agents sweep them up and sift for gold. In four months, agents contacted 130,000 leads and created 3,200 opportunities."

That is a 2.5% lead-to-opportunity rate. It is a modest number, and that is exactly why it is worth quoting — it is what a real deployment looks like when someone reports it honestly rather than selling it. The value was not a miracle conversion rate. It was that 130,000 leads went from zero touches to some touches.

For context on how new this is: The Bridge Group's 2025 SDR report, surveying 351 B2B companies, recorded "AI SDRs" as a distinct category for the first time in 2025 — at 1% of respondents. The same report found SDR quota attainment at 60%, the lowest in the study's history, with median tenure of 1.9 years and 40% annual attrition. The pressure is real; the adoption is early. Anything you read claiming 75% market adoption is invented.

Broader adoption, from Salesforce's data: 54% of sales teams now use AI agents in some form, with another 34% expecting to within two years. Among teams already using them, 94% of sales leaders call them critical to meeting business demands.

A response-time SLA you can actually run

Speed is an operating decision, not a tooling decision. Here is a target set that follows the decay curve rather than a round number someone picked.

Window

What must happen

Why

0–2 min

Qualified response: the visitor's actual question answered from real business data, and at least one qualifying detail captured

This is where the market has split. 47% of responders in the RevenueHero test were here

0–5 min

Lead scored and routed to a named owner with context attached

The 5→10 minute decay is 4x on qualification odds

Within 1 hour

Human contact for anything scored hot

HBR: contact within an hour is ~7x more likely to qualify than an hour later

Same working day

Human contact for warm

Beyond 24 hours, HBR's data shows the drop-off exceeds 60x

Never

An auto-reply counted as a response in your reporting

It is the most common way teams fool themselves

Two rules matter more than the table:

Measure time-to-qualified-response, not time-to-first-response. If your reporting cannot tell the difference between an autoresponder and a real answer, your reporting is flattering you.

Give every conversation a human exit. Not buried, not after five failed attempts. 89% of consumers expect it, and the one product metric worth chasing in this category is that 15% seamless-handoff figure.

How ConGreeto handles this

ConGreeto is an AI sales assistant built specifically around time-to-qualified-response rather than time-to-first-response.

It answers from your real business. Your site, documents and listings are crawled and indexed, and answers are retrieved from that content at answer time — hybrid semantic plus keyword retrieval — so the assistant quotes your delivery window and your availability rather than inventing them. The "wrong opening hours for months" failure is a grounding failure, and grounding is an architecture decision.

It captures as it goes. Contact details and qualifying fields are saved during the conversation as they surface, not gated behind a form at the end. A visitor who leaves halfway through still leaves you something to work with.

It scores and routes. Every completed conversation produces a summary, the captured fields, and a tier — Hot, Warm, Nurture or Cold — judged against that industry's conversation norms, landing on a pipeline board with an owner and a follow-up rather than in an inbox.

It hands off with context attached. The transcript and the captured fields travel with the lead, so the human who picks it up does not re-ask what was already answered.

It is honest about what it is. It is an assistant that qualifies and routes, not a synthetic salesperson. Given the Marketing Science disclosure finding, that is a deliberate trade: we would rather lose the uplift that comes from deception than build a product whose value depends on not being noticed.

It works in over 30 languages including right-to-left, detecting the visitor's language rather than making them find a switcher — which matters more than it sounds, because your response time for an enquiry in a language nobody on your team reads is currently infinite.

If you want the engineering behind that, our parent company wrote it up in Why 40% of AI Agent Projects Get Cancelled, and the product-side reasoning is in AI Chatbot for Lead Generation: Why We Built ConGreeto.

Plans start at $37.50/month billed annually. Installation is one script tag.

Start with ConGreeto →

Frequently asked questions

What is a good lead response time? Under five minutes is the widely-cited target, and the underlying research supports urgency — but the decay curve is steepest at the very start. The 2007 InsideSales/Oldroyd data found qualification odds fall roughly 4x between minute five and minute ten. Treat five minutes as the outer limit rather than the goal, and measure whether the response actually answered anything.

What is the 5-minute rule in sales? It comes from research by Dr. James Oldroyd presented in October 2007, analysing over 15,000 leads and 100,000 call attempts across six companies. It found that the odds of contacting a lead fall 100x, and the odds of qualifying one fall 21x, between a 5-minute and a 30-minute response. Importantly, the study explicitly did not measure close rates.

What is the average lead response time in 2026? There is no credible 2026 study. The widely-quoted "42 hours" is from Harvard Business Review in 2011. The most recent real measurement we could find is RevenueHero's March 2024 test of 1,000 B2B SaaS companies: 63.5% never responded, and those that did averaged one day, five hours and seventeen minutes. Be sceptical of any page presenting a precise 2026 figure without a stated sample size and method.

What percentage of companies never respond to leads at all? 23% in HBR's 2011 audit of 2,241 companies. 63.5% in RevenueHero's 2024 test of 1,000 B2B SaaS companies — and that test counted automated replies as a response. In real estate, WAV Group's 2014 mystery shop of 384 brokers found 48% of enquiries were never answered.

Is an automated reply the same as a lead response? No, and conflating the two is the most common way sales teams overstate their performance. An autoresponder scores perfectly on time-to-first-response and changes nothing. Measure time-to-qualified-response: a reply that answers the actual question, captures new information, states what happens next, and can reach a human.

How do you measure lead response time correctly? Timestamp the enquiry and timestamp the first substantive reply, excluding automated acknowledgements. Report the distribution, not the average — averages hide the bimodal split between instant responders and non-responders. Segment by arrival hour, because after-hours performance is usually where the failure is concentrated.

What percentage of enquiries arrive outside business hours? We could not find a credible primary study behind any of the commonly-quoted figures, and we would treat them all as unsourced. The defensible data is behavioural: Jobber's 2026 report found 41% of jobs booked online arrive after hours, and CallRail's 2025 survey found 82% of consumers will call a competitor when a call goes unanswered, while only 42% leave a voicemail.

Can a chatbot replace a sales rep for first response? For the first response, often yes — for the sale, no. The evidence supports a narrower claim: an assistant that answers accurately from real business data, captures qualifying details and routes to a human quickly. Consumer research is consistently hostile to AI-only service: 79% strongly prefer human interaction and 89% expect a human option to exist.

Does disclosing that a chatbot is AI hurt conversion? Yes, measurably. A randomised field experiment on 6,200+ customers published in Marketing Science (2019) found disclosure cut purchase rates by 79.7%. We disclose anyway, and we would advise you to. A product whose value depends on customers not realising what they are talking to is a liability rather than an asset.

What is a realistic lead response SLA for a small team? A qualified automated response within two minutes at any hour; scored and routed within five; human contact within one hour for hot leads and the same working day for warm. The point of automating the first two rows is that the third and fourth become achievable without a night shift.

How do you score a lead as hot, warm, or cold? Against your industry's own conversion norms rather than a generic template — the signals that predict a sale for a dental practice are not the ones that predict a sale for a car dealership. What makes a score trustworthy is that it is derived from the conversation itself (what they asked for, timeline, budget, specificity) and is auditable back to the transcript.

Sources & further reading

From us

ConGreeto is built by Brillnex Systems, a software and AI engineering firm building production AI systems for startups and growing businesses.