B2B Cost Per Lead Benchmarks 2026: $84 to $3,080

Look up what a B2B lead costs and the answer comes back as $84. Or $213. Or $237. Or somewhere between $420 and $3,080. All four are in circulation right now, all four are labelled cost per lead, and the gap between the cheapest and the dearest is roughly a hundred to one.

A hundred to one is not a market difference. No category of spending varies that much between competent and incompetent teams. It is what happens when raw form fills, marketing-qualified leads, sales-qualified leads and booked meetings are all reported under one word, by publishers who mostly do not say which one they counted.

This page does not add another set of medians to that pile. It audits the ones already there. Every figure below is published with the stage it counts, the sample it rests on, the window the data was collected in, and whether the attribution survives being checked. Two of the most-quoted numbers in this category do not survive it, and one of those is the figure most people reach for first.

What B2B cost-per-lead figures actually disagree about

Cost per lead is the total spend on a channel divided by the number of leads that channel produced. The arithmetic is trivial. Everything difficult lives in the second half of that sentence, because “lead” is not a defined unit.

Two teams can run that calculation honestly on the same quarter of the same business and report $60 and $2,000. Neither has made an error. One divided spend by every contact record created. The other divided the same spend by the handful of those records a salesperson agreed to work. The published benchmarks do exactly this to each other, at scale, and rarely say so.

Here is what the major 2026 figures actually measure, side by side.

Read the third column before the first. Belkins and HubSpot are not disagreeing about the price of a lead by a factor of thirty. They are answering two different questions, and only one of them said which.

The five things publishers call “a lead”

A lead is any of five distinct units, and the cost changes by an order of magnitude depending on which one a publisher counted. This is the single most useful table on this page.

Move a single figure across that table and it changes more than any optimisation you could run on it. A channel at $60 per inquiry with a 20% inquiry-to-MQL rate and a 15% MQL-to-SQL rate costs $2,000 per SQL. Same spend, same channel, same quarter. The only thing that changed is which row you counted.

Where the $84 top-quartile figure came from

One number dominates this category. It appears across dozens of 2026 pages in the same shape: top-quartile B2B programmes run a cost per lead of $84, the median is $213, and the bottom quartile is $397, giving a 4.7x spread between disciplined and undisciplined teams. It is usually credited to HubSpot’s State of Marketing 2026, sometimes to a joint HubSpot and Demand Gen Report study covering more than 8,500 companies globally.

Checked against the sources, that account does not hold.

HubSpot does publish an $84 figure. On its CPL and CAC benchmarks page, last updated 14 July 2026, the sentence reads: “Average B2B CPL across all channels is $84.” It is an average. The word quartile does not appear on that page. Neither does median, nor $397, nor $213, nor 8,500, nor any reference to Demand Gen Report.

HubSpot’s State of Marketing 2026 report, meanwhile, publishes no cost-per-lead benchmarks at all. And the joint 8,500-company study could not be located anywhere on the public record. None of the pages carrying that attribution links to it, and several of them list twenty numbered sources without including the study they credit for their headline number.

So the only figure in that set traceable to a named publisher is $84, and on that publisher’s own page it is a central average rather than the top of a distribution. The median and the bottom quartile wrapped around it, and the sample they are attributed to, have no locatable source. A single average appears to have been re-badged as a quartile boundary, and a distribution constructed around it after the fact.

Four independent sources, one dataset

The second problem is quieter and affects more pages. Independent corroboration is the usual test for a benchmark: if four publishers report similar industry figures, the figures are probably sound. In this category that test fails, because the four are not independent.

First Page Sage’s cost-per-lead-by-industry report is the origin. It gives B2B SaaS a blended CPL of $237, cybersecurity $406, financial services $653, software development $591, manufacturing $553. The data was collected between January 2022 and June 2025. The report does not disclose a sample size.

HubSpot’s industry table carries those same values and states plainly that it “combines insights from FirstPageSage’s 2026 report”, despite the page being badged as HubSpot Research. Prospeo publishes the same table and credits First Page Sage explicitly, naming the same collection window. SalesHive carries the same $237 and $198 figures without naming the origin.

Four tables, one dataset, collected in a window that closed in June 2025 and presented throughout as a 2026 benchmark. Averaging them does not strengthen the number. It counts one unsampled table four times.

B2B cost per lead by channel, with denominators attached

Channel figures are the most requested and the worst documented. Below is what the major publishers report, with the denominator each one uses and what is disclosed about its method. These are not IVRIS measurements. IVRIS has not run a CPL study, and a table presented as ours would be exactly the problem this page is about.

Two structural warnings about this table. Webinar and event rows mix registrations, attendees, badge scans and booked meetings, which is why their ranges look wilder than their true cost variance. And both ABM rows count accounts rather than people, so they belong to a different denominator entirely and should never be averaged with the rest.

Cost per lead by industry, and what those tables inherit

Industry figures come from two sources with incompatible denominators, and the difference between them is the clearest illustration on this page of why a single benchmark cannot exist.

First Page Sage’s column is blended paid and organic, counts raw leads, was collected between January 2022 and June 2025, and carries no published sample size. Belkins’ column comes from agency delivery data across more than 1,000 companies in 50-plus industries, and its methodology states that CPL is “measured at the sales-qualified stage rather than initial capture”. Belkins is the more transparent of the two about what it counts and the less transparent about when.

The consistent three-to-five-fold gap between the columns is not a disagreement. It is the price of qualification, visible for once because one publisher declared its stage.

This is a B2B page

Every figure above is B2B. Local and consumer services run on different economics entirely, with far lower absolute costs and a much shorter path from enquiry to job. If you are pricing leads for a home-services or local business, the channel-by-channel figures in our guide to local lead generation are the right reference, and mixing the two sets will mislead you in both directions.

Why the cheapest channel is often the most expensive

The practical consequence of all this is a ranking problem. Channels are almost always compared on cost per inquiry, because that is the number every ad platform reports without being asked. But qualification rates vary more between channels than costs do, and the two tend to move in opposite directions.

Take two channels. The first delivers inquiries at $60, converts 20% of them to MQL and 15% of those to SQL. The second costs $300 per inquiry, converts 60% to MQL and 40% of those to SQL.

Channel A looks five times cheaper and is 60% more expensive for the only thing either channel exists to produce. This is not a contrived example. Self-serve and broad-reach channels routinely collect large volumes of contacts with weak intent, while high-cost channels tend to be aimed at people who already qualify.

The multiplier in the last row is the number worth tracking. Channels with a low multiplier convert cost into pipeline efficiently regardless of what they charge per inquiry. Anything above about 15x is worth investigating before it is worth scaling.

To run this on your own numbers you need each channel’s own conversion rates, not a company-wide average. A blended rate cancels out precisely the difference the calculation is meant to expose. Our B2B lead conversion rate benchmarks covers what the published MQL-to-SQL rates do and do not measure, and why the widely quoted 13% is a different metric altogether. Use it for context, and your CRM for the rates themselves.

Cost per lead is also the input to the metric your finance team actually cares about. If cost per SQL is where marketing efficiency shows up, customer acquisition cost is where it lands, and the relationship between them is set by the win rate you apply after this calculation ends.

How to work out your own cost per qualified lead

Run it once and the ranking usually reorders. Run it quarterly and you can see qualification rates drift, which is the earliest warning that a channel is decaying while its cost per inquiry still looks healthy.

Where the public record runs out

An honest benchmark page should say what it could not establish. Four things.

No B2B cost-per-lead figure in circulation has both a disclosed sample size and a stated qualification stage. Belkins gives the stage and the company count but no collection window. First Page Sage gives the window but no sample. Everyone else gives neither. That combination does not currently exist in public, which is why this page publishes no IVRIS median.

Nobody has published cost per lead segmented by qualification stage from the same population. The three-to-five-fold gap between raw-lead and SQL-stage tables is inferred by comparing two different publishers’ datasets. A single organisation reporting both from one book of business would settle the question and does not appear to exist.

The quartile distribution everyone quotes has never been measured. There may well be a 4.7x spread between disciplined and undisciplined B2B programmes. It is plausible. It is simply not something any locatable study has established.

Currency is almost never stated. Every figure on this page is assumed to be US dollars because that is the convention, not because the publishers said so. For teams buying leads in other markets, that assumption is doing more work than it should.

Methodology, sources and revision history

Each figure was opened at its own URL and read for publisher, publication or update date, collection window, sample size, denominator, currency and paid-or-blended basis. Where a page attributed a figure to a study, that study was searched for separately rather than taken on trust. “Untraceable” throughout means the named study could not be located on the public record as at 6 August 2026; it is not a claim that no such study exists anywhere.

IVRIS has not conducted a cost-per-lead study and publishes no CPL benchmark of its own on this page. The worked examples use round illustrative figures, clearly labelled as such, and are not drawn from client data. Where this page reports a range, the range is the span of what other publishers report, not a measurement.

Primary sources: HubSpot CPL and CAC benchmarks (page updated 14 July 2026); First Page Sage, Average Cost Per Lead by Industry, published 8 May 2025 and updated 23 December 2025, data collected January 2022 to June 2025; Belkins, B2B Cost Per Lead Benchmarks, 20 March 2026; Zeliq, 29 December 2025; SalesHive, updated 10 July 2026; Google AI Overview for “cost per lead benchmarks”, captured 2 August 2026.

Version 1.0, 6 August 2026. Figures verified 6 August 2026. This page is on a 90-day review cycle because paid-channel costs and AI Overview output both move faster than the annual benchmark reports they are drawn from.

Frequently Asked Questions