The cost per lead formula is one division, and it is the least problematic part of the subject. You divide a spend by a number of leads. Everything separating two companies that report the same figure comes from what each agreed to count in the denominator.

None of the pages publishing cost per lead tables by industry defines what a lead is before counting them. This page starts there.

A second problem sits underneath, and it is more rarely stated. The phrase describes two different economic objects: the price of a contact bought from a supplier, and the internal cost recorded in your own advertising account. Comparison tables mix the two without saying so.

The word “lead” has a standard definition, and it is not yours

This is the underlying fact, and it explains everything that follows.

What is standardised in advertising. The viewable ad impression has an accredited measurement standard: for display, at least 50% of the ad’s pixels in the foreground tab for at least one continuous second, dropping to 30% of pixels for large formats at or above 242,500 pixels, and for video 50% of pixels with two continuous seconds of play. Some forty companies are accredited on that measurement. Two platforms reporting a viewable impression are talking about the same thing.

What exists for leads, and nobody quotes. The Media Rating Council, the body that accredits advertising measurement, does publish a definition, in its September 2022 outcomes standards. The section heading tells you where this is going: “Leads (Not a Direct Measure of Sales)”.

What it says exactly. “Leads represent potential consumers who may have a need for or interest in a product/category or otherwise be targeted for advertising, but who have not provided any direct indication or communication that signifies purchase intent for a specific product.”

In other words, that is not what you call a lead. For the MRC, a lead is a targeting audience, not a person who filled in a form. That person falls under a different section of the same standards, covering inquiries and requests for information.

And the text goes further. “If a campaign goal is to generate interest in a specific product or to promote action, leads should not be used as a KPI for success or in determining incrementality.” The body that standardises advertising measurement explicitly advises against managing on this number.

What it recommends instead. Qualified leads, which it says “are required to be used as a measure of campaign success” when that is the campaign goal, with a requirement almost nobody meets: “The basis of qualification must be disclosed, supported and demonstrable.”

The real finding, then. It is not that no standard exists. It is that one exists, that it names something other than the commercial use of the word, that it advises against managing on the raw figure, and that no market participant is accredited on the measurement at all.

What the Media Rating Council actually standardises for impressions and for leadsDiagram comparing what the Media Rating Council actually standardises for the viewable advertising impression and for the lead. The viewable ad impression carries an accredited measurement standard requiring, for display advertising, at least fifty percent of the advertisement’s pixels in the foreground tab for at least one continuous second, dropping to thirty percent of pixels for large formats at or above two hundred and forty-two thousand five hundred pixels, and for video at least fifty percent of pixels together with two continuous seconds of play, with some forty companies accredited on that measurement, so two platforms reporting a viewable impression are describing the same thing. The same body also publishes a definition of leads, in its September two thousand and twenty-two outcomes standards, under a heading stating that leads are not a direct measure of sales. That definition describes potential consumers who may have a need for or interest in a product or category, or otherwise be targeted for advertising, but who have not provided any direct indication or communication signifying purchase intent for a specific product, which describes a targeting audience rather than a person who completed a form, such a person falling under a separate section of the same standards covering inquiries and requests for information. The text further states that where a campaign goal is to generate interest in a specific product or to promote action, leads should not be used as a key performance indicator for success or in determining incrementality, and it requires that the basis of qualification for a qualified lead be disclosed, supported and demonstrable. No market participant is accredited on lead measurement.One unit is standardised. The other is standardised as something else.The viewable impression50% of pixels, 1 continuous second30% for formats above 242,500 pxVideo: 50% of pixels, 2 secondsIndependent audit~40 companies accredited.The lead, as standardisedA targeting audienceNo demonstrated purchase intent”Not a Direct Measure of Sales”Nobody is accredited on itThat is not your lead.The standard exists, it names something else, and it advises against managing on the raw figure.Someone who fills in a form falls under a different section: inquiries and requests for information.
The standard for leads exists. It names a targeting audience with no demonstrated purchase intent, and advises against using it as a success KPI. Source : Media Rating Council (2022)

What the platforms count instead

They are not measuring what the standards body describes either, and their own definitions say so.

Google’s definition of a conversion. A conversion occurs when an interaction with your ad “leads directly to a behavior that’s valuable to you”, and Google gives four examples at the same rank: a purchase, a newsletter sign-up, a phone call, a download. Commercial qualification appears nowhere in the definition. To the platform, a downloaded PDF and a quote request are the same object.

What LinkedIn publishes about its own lead product. The Lead Gen Forms page states that “90% of pilot customers beat their cost-per-lead (CPL) goals, with lower CPLs compared with their standard Sponsored Content campaigns.” There is no methodology, no period, no number of pilot customers, and no definition of what the goal was. It is a marketing claim, and it should be cited as one.

The immediate consequence. The denominator of your cost per lead is defined by whoever calculates it. One company counting every document download, another counting only demo requests, and a third requiring a valid phone number will publish three unrelated numbers under the same word.

Why that invalidates the industry tables. Comparing your cost per lead to a sector average assumes the sector counts the way you do. Nothing guarantees it, nothing verifies it, and none of the published tables states what was counted.

The first move, before any arbitration. Write the definition down. What counts as a lead, what counts as a qualified lead, what counts as an opportunity, who decides the transition between them, and within what deadline. Without that document, no figure on this page is manageable.

Two costs per lead that have nothing to do with each other

The distinction is old and it appears in glossaries. It has disappeared from the pages that publish numbers.

The first object: a billing model. You pay a supplier per contact delivered. The price is negotiated, contractual, and it contains that supplier’s margin along with their own acquisition cost. The useful question is then about the quality delivered, the freshness and the exclusivity of the contact.

The second object: an internal indicator. You record, in your own ad account, what a form cost you. The figure is an auction output you do not set, and it includes neither your team’s time, nor your tooling, nor content production.

Why the two cannot be compared. The first contains a commercial margin the second does not have. The second omits costs the first has already absorbed. Two identical amounts describe two different economies.

The error this produces in a meeting. People negotiate a figure that cannot be negotiated, or optimise a price that should be contracted. Anyone comparing a supplier’s asking price to the cost shown in their ad account concludes wrongly almost every time.

What to ask before commenting on any cost per lead. Is this a price paid to somebody, or a cost recorded here? The answer changes everything that follows, and it is almost never volunteered.

Cost per lead as a purchase price and as an internal indicatorDiagram contrasting the two economic objects that the phrase cost per lead covers. The first is a billing model in which an advertiser pays a supplier for each contact delivered, the price being negotiated and contractual and containing that supplier’s margin as well as their own acquisition cost, so the useful question concerns the quality delivered, the freshness and the exclusivity of the contact. The second is an internal indicator in which the advertiser records, in their own advertising account, what a form submission cost, this figure being an auction output the advertiser does not set and including neither the team’s time, nor the tooling, nor content production. The two cannot be compared because the first contains a commercial margin the second does not have, while the second omits costs the first has already absorbed, so two identical amounts describe two different economies. The error this confusion produces is that people negotiate a figure that cannot be negotiated or optimise a price that should be contracted, anyone comparing a supplier’s asking price to the cost shown in their own advertising account concluding wrongly almost every time. The question to ask before commenting on a cost per lead is therefore whether the figure is a price paid to somebody or a cost recorded internally.The same word, two economiesA purchase priceYou pay a supplier per contact delivered.Negotiated, contractual.Contains the supplier’s marginand their own acquisition cost.The useful question: quality, freshness,exclusivity of the contact.An internal indicatorWhat a form submission cost you.An auction output you do not set.Excludes your team’s time, your toolsand content production.The useful question: what becomes ofthose contacts after the form.The question to ask: is this a price paid to somebody, or a cost recorded here?The answer changes everything that follows, and it is almost never volunteered.
Two identical amounts describe two different economies. One contains a commercial margin, the other omits internal costs. Source : MASTRATOS (2026)

Where the industry tables actually come from

The chain is short, and shorter than it looks.

The table that irrigates the subject. First Page Sage publishes cost per lead by industry, blended, paid and organic: B2B SaaS at $237 blended, legal services at $649, financial services at $653, higher education at $982, and so on across thirty rows.

Its methodology, quoted in full. “Our 2026 report on Cost Per Lead by Industry was prepared by our marketing research team, based on data collected between January 2022 through June 2025.”

That is the whole of it. No sample size, no geography, no statement of whether these are means or medians, no definition of what was counted as a lead. “Our dataset” is never defined.

The circularity that makes two sources into one. HubSpot, the largest marketing publisher quoting CPL benchmarks, states on its own page that “the following data on the average cost per lead by industry combines insights from FirstPageSage’s 2026 report.” Citing HubSpot for a 2026 sector CPL is citing First Page Sage a second time.

The figure that will not die. $198.44, still quoted as an average cost per lead, comes from a HubSpot demand generation report published in 2017. It was a declarative survey: respondents chose from bands with an open upper bound of $1,001 and above, 18% answered that they did not know, and sector sub-samples ran from 15 to 97 respondents. Two decimal places on a mean of declared bands is precision that does not exist.

The one benchmark resting on measured data, and its limit. WordStream reports $66.69 cost per lead across industries, on a stated sample of 13,474 US search campaigns running April 2025 to March 2026, and states plainly that its “averages” are technically median figures. It is paid search only, so it is a cost per conversion in one channel, not a blended cost per lead. Note also that the same numbers are republished on a more accessible page with the methodology block removed.

Real provenance of the published cost per lead benchmarksTable setting out the real provenance of the cost per lead benchmarks published in English. The first is a cost per lead by industry report giving blended, paid and organic figures across around thirty rows, for instance two hundred and thirty-seven dollars blended for business-to-business software as a service, six hundred and forty-nine dollars for legal services, six hundred and fifty-three dollars for financial services and nine hundred and eighty-two dollars for higher education, whose entire published methodology is a single sentence stating that the report was prepared by the publisher’s marketing research team based on data collected between January two thousand and twenty-two and June two thousand and twenty-five, with no sample size, no geography, no statement of whether the figures are means or medians and no definition of what was counted as a lead. The second apparent source is the largest marketing publisher quoting cost per lead benchmarks, which states on its own page that its data combines insights from that same report, so citing it amounts to citing the first source a second time. The third is a figure of one hundred and ninety-eight dollars and forty-four cents still quoted as an average cost per lead, which comes from a demand generation report published in two thousand and seventeen based on a declarative survey whose respondents chose from bands with an open upper bound of one thousand and one dollars and above, in which eighteen percent answered that they did not know and sector sub-samples ran from fifteen to ninety-seven respondents. The fourth rests on measured platform data, reporting sixty-six dollars and sixty-nine cents across industries on a stated sample of thirteen thousand four hundred and seventy-four United States search campaigns running from April two thousand and twenty-five to March two thousand and twenty-six, and stating that its averages are technically median figures, but it covers paid search only and is therefore a cost per conversion in one channel rather than a blended cost per lead.Four benchmarks, and what is behind themThe industry tableB2B SaaS $237, legal $649, higher ed $982Methodology in full: one sentence. No sample, no geography, no definition.The large publisher quoting itLooks like a second sourceStates on its own page that it combines insights from the same report.The $198.44 averageStill quoted as current2017 survey, banded answers, 18% said they did not know.The measured one$66.69, 13,474 US campaignsDeclares its sample and that its “averages” are medians. Paid search only.The only one you can audit is the one that covers a single channel, so it is not a blended cost per lead.
Two apparent sources, one dataset. And the figure still quoted as an average comes from a 2017 declarative survey with banded answers. Source : MASTRATOS from First Page Sage, HubSpot, WordStream (2026)

Three ways to lower a cost per lead without improving anything

All three work immediately, and none produces one more meeting.

Broaden the definition of a lead. Add document downloads, newsletter sign-ups and pricing page visits to the counter. The denominator inflates, cost per lead collapses, and the number of sales conversations has not moved.

Remove a form field. Completion rate rises, cost per lead falls, and your sales team now spends time qualifying by phone what the form used to qualify by itself. The cost moved, it did not disappear.

Shift budget toward a less qualified segment. Broad audiences complete a form more cheaply than in-market audiences. Your cost per lead improves while your cost per closed deal degrades.

What makes these hard to spot. None is a mistake. Simplifying a form and broadening an audience are good practices in other contexts. They simply lower this ratio for reasons unrelated to commercial performance.

The question that exposes all three. How many of those leads produced a meeting that actually happened? If cost per lead falls while meetings stay flat, you broadened the denominator, you did not improve acquisition.

Three methods that lower a cost per lead without real improvementDiagram setting out three methods that lower a cost per lead immediately without producing one additional sales meeting. The first is broadening the definition of a lead by adding document downloads, newsletter sign-ups and pricing page visits to the counter, which inflates the denominator and collapses the cost per lead while the number of sales conversations has not moved. The second is removing a field from the form, which raises the completion rate and lowers the cost per lead while transferring to the sales team the qualification work the form used to perform, so the cost is moved rather than removed. The third is shifting budget toward a less qualified segment, since broad audiences complete a form more cheaply than in-market audiences, so the cost per lead improves while the cost per closed deal degrades. These three cases are hard to spot because none of them is a mistake, simplifying a form and broadening an audience being good practices in other contexts, and they simply lower this ratio for reasons unrelated to commercial performance. The question that exposes all three is how many of those leads produced a meeting that actually happened, since a cost per lead falling while meetings stay flat signals a broadened denominator rather than improved acquisition.Falling without improvingBroaden the definition of a leadThe denominator inflatesNo new enquiriesDownloads, newsletter sign-ups, pricing page visits.Remove a form fieldCompletion risesThe cost movesSales now qualify by phone what the form qualified by itself.Aim widerCPL improvesCost per deal risesBroad audiences fill a form more cheaply than in-market audiences.The question that exposes them: how many produced a meeting that actually happened?None of the three is a mistake. That is what makes them hard to catch in a report.
None of the three is a mistake. The question that exposes them: how many of those leads produced a meeting that actually happened? Source : MASTRATOS (2026)

Raw and qualified: the gap the platform documents itself

This is the strongest argument in the file, and it comes from Meta.

What Meta publishes about its automated lead campaigns. Raw cost per lead falls by 14%, at more than 95% confidence. The result is clean.

What it becomes on the qualified lead. The improvement drops to 10%, and crucially to 83% confidence, below the usual 90% threshold, across nineteen tests run from November 2024 to January 2025.

What that establishes. The two curves do not move together. On the same setup, the same period and the same sample, the gain on the raw figure is demonstrated and the gain on the qualified one is not. The platform publishes both numbers itself.

What to take from it for management. Tracking raw cost per lead means rewarding whichever setting produces volume. That is exactly what the algorithm does when you teach it nothing else.

In B2B, this is the point that costs money. An unqualified lead is not neutral, it consumes sales time. A cost per lead falling 14% while the qualification workload rises is a degradation presented as progress. The full figures and their limits are in our analysis of Meta Advantage+.

Gap between the gain on raw cost per lead and on qualified cost per leadChart comparing two results published by Meta for its automated lead generation campaigns, measured across nineteen tests run from November two thousand and twenty-four to January two thousand and twenty-five. Raw cost per lead falls by fourteen percent at more than ninety-five percent confidence, which is a clean result. On qualified cost per lead the improvement drops to ten percent and, more importantly, to eighty-three percent confidence, which sits below the usual ninety percent threshold and therefore means the gain is not demonstrated. On the same setup, the same period and the same sample, the gain on the raw figure is established and the gain on the qualified figure is not, and the platform publishes both numbers itself. It follows that tracking raw cost per lead amounts to rewarding whichever setting produces volume, which is exactly what the algorithm does when it is taught nothing else. In business-to-business advertising this is the point that costs money, since an unqualified lead is not neutral but consumes sales time, so a cost per lead falling fourteen percent while the qualification workload rises constitutes a degradation presented as progress.Raw and qualified do not move togetherAutomated lead campaigns, 19 tests, November 2024 to January 2025.Raw cost per lead-14%at more than 95% confidenceQualified cost per lead-10%at 83% confidence83% sits below the usual 90% threshold. The gain on qualified leads is not demonstrated.Same setup, same period, same sample. The platform publishes both numbers itself.
Meta publishes both numbers. The second falls below the usual significance threshold. The two curves do not move together. Source : Meta (2025)

Making cost per lead manageable: send the qualification back

There is only one technical way out, and it is well known.

The problem to solve. The advertising platform only learns what you send it. If you send “form submitted”, it optimises the number of form submissions, quite logically.

What you need to send instead. The contact’s status after sales contact: lead accepted, meeting held, opportunity opened, deal signed. Those are the events that should flow back, not the form.

What it is called. Offline conversion import, which Google now recommends evolving into enhanced conversions for leads, built on user-provided data.

The prerequisite nobody mentions. Your CRM has to carry those statuses, somebody has to maintain them, and the rule for moving between them has to be written down. Without that, there is nothing to send back.

The honest limit. This assumes enough volume for the platform to learn from. At a few dozen leads a month it documents your own management without feeding the algorithm. That is still useful, but do not expect an automatic gain.

What it also fixes. The reconciliation between platform counters and your CRM, which we cover in why GA4 and Meta conversions don’t match and in server-side tracking.

From CPL to CAC: the only chain that gives the number meaning

A cost per lead is not a destination. It is a step, and it only has value relative to what follows.

The calculation that puts things back in order. A $40 cost per lead at a 10% close rate gives a $400 customer acquisition cost. The same cost per lead at 2% gives $2,000. The first number did not change.

What that means for an arbitration. Comparing two channels on cost per lead is betting their close rates are identical. They almost never are, and the gap between them is often larger than the gap in cost per lead.

The most frequent case in B2B. One channel produces leads at half the cost and closes at a third of the rate. It wins on cost per lead and loses on acquisition cost, and a dashboard tracking only the first points you the wrong way for quarters.

What to calculate instead. Cost per lead, multiplied by the inverse of your close rate, compared with your margin per deal. That chain is developed in how to calculate CAC and in ROAS vs MER vs CAC vs LTV.

Buying leads is a different business, and a different contract

Paying a network per delivered contact is not managing your cost per lead. It is buying an asset, and it should be contracted as one.

What has to be in the contract. Exclusivity, or the number of buyers the same contact is sold to. Freshness, meaning the delay between the contact’s action and its delivery to you. The qualification criteria applied before delivery, and who applies them. And the replacement terms for a contact that turns out to be unreachable or out of scope.

Why the unit price says nothing on its own. A contact sold exclusively, delivered within the hour, filtered on company size, is a different product from the same contact sold to four buyers a week later. Comparing their prices is comparing two different goods.

The measurement that settles it. Not the price per lead, but the cost per meeting held. On bought leads that ratio is usually far worse than the price suggests, and it is the only figure that lets you compare buying with generating.

From what volume does your cost per lead mean anything?

The question precedes any interpretation, and it is rarely asked.

The principle. A monthly cost per lead computed on a small number of leads moves more by chance than by decision. A few leads more or less shift the figure by tens of percent with no cause behind it.

The order of magnitude in B2B. Below a few dozen leads a month, the month-to-month variation is not interpretable. On five or ten leads, it never is.

What to look at instead. A three-month moving average, and cost per qualified opportunity rather than per lead. The second is steadier because it removes the most volatile part of the denominator.

The rule. Never comment on a cost per lead without showing the number of leads behind it next to it. That is the only information that says whether the gap deserves a discussion at all.

Questions to ask whoever shows you a cost per lead

What is the denominator? What exactly was counted as a lead this month, and is it the same thing as last month. If the answer is not immediate, the rest of the discussion has no object.

Is this a price paid or a cost recorded? The first contains a supplier’s margin, the second omits your time. Not the same dollars.

What spend is in the numerator? Media only, or with fees, tools and content production. A media cost per lead is useful for arbitrating a campaign, it says nothing about real cost.

How many of those leads produced a meeting that happened? This is the question that separates an improvement from a broadened denominator, and the hardest to answer without a maintained CRM.

And what does a signed deal cost over the same period? The only question whose answer lets you decide anything.

Where to go next

You want the cost of a customer, not a contact. How to calculate CAC.

You want to know which metric to run on. ROAS vs MER vs CAC vs LTV.

Your counters disagree. Why GA4 and Meta conversions don’t match.

You are structuring a search account. Google Ads account structure for B2B.

Your click-through rate is moving and you do not know if it is real. Click-through rate benchmarks.

Your landing page is where the leads die. What makes a B2B landing page convert.

In short

  • The formula is trivial, the denominator is not. A standard does define a lead, but it names a targeting audience with no demonstrated purchase intent, and advises against using it as a success KPI.
  • Two objects share the name: a price paid to a supplier, which contains their margin, and a cost recorded here, which omits your time.
  • The platforms are not measuring qualification either. Google’s definition of a conversion puts a download and a quote request at the same rank.
  • Two apparent benchmark sources are one dataset. The largest publisher quoting CPL states on its own page that it combines insights from the industry table it appears to corroborate.
  • The $198.44 average is a 2017 declarative survey with banded answers and 18% of respondents who did not know.
  • Three methods lower a CPL without improving anything: broaden the definition, lighten the form, aim wider.
  • Meta publishes that the gain on raw leads is demonstrated and the gain on qualified leads is not.
  • A CPL only means something multiplied by a close rate and compared with your margin per deal.

A cost per lead is not compared to a benchmark, it is defined before it is measured. Book a diagnostic, or see how we approach B2B paid acquisition.