Welcome back, jump to Tutorials, Support & Docs, or What’s New. Go to PlatformNot you?Log out
Platform AI Agents Customers Developer Docs Services Company Careers Blog Resources Comparisons Buyer's Guide Integrations Use Cases Tutorials Events Sign In — Classic view Sign In — AI agency Book a Demo
2026 B2B Benchmark · Insight 03 of 08

Three channels sold a lead for under $210. Only one shows what a customer cost.

The cheap-lead illusion.

Cheap leads were everywhere in metadata.io’s 2025 B2B benchmark, Instagram at $138, Facebook at $145, LinkedIn at $202 (49, 71 and 138 advertisers). What a customer cost is the number that decides a budget, and only one channel here can show it: LinkedIn’s leads became customers at $63,312 each and returned 0.57x. 57 cents back on every ad dollar, counting only revenue we can trace to an ad. Across the whole dataset a customer cost $58,887 and a dollar came back as 0.56x, so LinkedIn is a rounding step from the pack rather than an escape from it. Why only one: on Facebook and on Instagram a single advertiser held too much of the closed-won revenue for the number to describe a market, so both are withheld (why).

Wire closed-won revenue back to the channel that sourced it this week, so cost per customer by channel is a standing report instead of a quarterly dig. You already have the cheap half, Facebook at $145 a lead, LinkedIn at $202. The half that decides budget is the half we mostly could not publish ourselves: what a Facebook or an Instagram customer cost is withheld (why), and LinkedIn is the one that made it through at $63,312 a customer and 0.57x back, across 114 advertisers we could follow into the CRM.

2025 · 34–138 advertisers behind each number · B2B · only groups with at least $50K spent · Benchmarks by channel Methodology · Suppression rules

Executive summary

  • The economics: LinkedIn’s $202 lead became a $63,312 customer, 0.57x back on the dollar, against $58,887 and 0.56x across the whole dataset. 21.2% of those deals closed, counting only the ones that actually closed or actually died.
  • The budget call: Rank channels on what a customer cost, not on what a lead cost, and if you cannot work out the customer number, treat the blank as the finding, not as a zero.
  • The limit: we watched 2025, we did not run an experiment, and on two of the three channels one advertiser held too much of the result, so what a Facebook or Instagram customer cost is withheld (why). That is a limit on what we can publish, not a verdict on either channel.

Playbook

  • The play: Rebuild the channel table with a cost-per-customer column next to the cost-per-lead column. LinkedIn is the one this dataset can price: $63,312 a customer, 0.57x back.
  • The setup: Carry channel and audience tags from capture through CRM to opportunity, wire it this week.
  • The measurement: A standing report that ranks channels twice, once by what a lead cost, once by what a customer cost. Any channel that moves two places between the two lists is the next budget conversation.

Which B2B ad channel is really cheapest?

It depends which question you ask, and which one you can actually answer. On what a lead cost in 2025, Instagram ($138, across 49 advertisers) and Facebook ($145) beat LinkedIn ($202), and Google Ads sold the dearest lead at $524 (56 advertisers). On what a customer cost, three of those four rows do not exist: when we checked on 2026-08-17, one advertiser held too much of the Facebook result and too much of the Instagram result for either to describe a market, and Google Ads had too few advertisers to publish at all. All three are withheld (why) rather than estimated.

That is the illusion in its plainest form. The number that is easy to produce exists for every channel; the number that decides a budget exists for one. And that one, LinkedIn at $63,312 a customer, paid back 0.57x, a rounding step from the 0.56x the whole dataset managed, on leads priced 39% above Facebook’s. Cheap leads did not buy a cheap customer here; they bought a customer nobody can see.

One set of channels, two questions, and we can only answer one of them for all four

Per lead: IG $138 · FB $145 · LI $202 · Google $524. Per customer: LI $63,312. Facebook, Instagram and Google Ads: withheld.

Channels ranked by what a typical lead cost, then switched to what a customer cost. 2025 lead-generation campaigns, and revenue we can trace back to an ad inside 12 months. Behind the bars: 49 to 138 advertisers on the lead numbers, 97 on the LinkedIn customer number. No one advertiser is allowed to swing a number, and where several ads touched a deal we split the credit evenly between them. The switch runs once as the chart comes into view; replay it with the buttons above. In the customer view only LinkedIn has a bar: Facebook, Instagram and Google Ads are drawn hatched, as withheld, never as zero. That the second view is nearly empty is the finding, not a broken chart. Colours as everywhere in this report: teal = the line in focus, orange = the cost contrast.

What did LinkedIn’s 3.1x click premium buy?

A 2.3x conversion premium, and then this dataset stops. LinkedIn’s clicks cost $11.90 against Facebook’s $3.80 in 2025 lead-gen campaigns, and 5.9% of LinkedIn clicks became leads versus 2.6% on Facebook. What the premium bought after the form fill is the sentence we cannot finish: the Facebook side of that comparison is withheld (why). On its own, LinkedIn’s premium ended at $63,312 per customer and 0.57x payback, better than nothing, still below one.

The annotation that matters: a ranking you can only work out for one of four channels is not a ranking. What a lead cost is here for all four; what a customer cost is here for LinkedIn alone, at $63,312 and 0.57x. Before you move budget on a cost-per-lead table, check whether the cost-per-customer table it is standing in for exists at all. On two of our three channels one advertiser held too much of the result, and we would rather say so than fill the gap in.
0.57x vs 0.56x
What came back per ad dollar, LinkedIn (114 advertisers) against every advertiser we could follow into the CRM (127), both under a dollar
$63,312 vs $58,887
Cost per customer, LinkedIn against the whole dataset, 2025. Facebook and Instagram: withheld
$11.90 vs $3.80
What a click cost, LinkedIn against Facebook, 2025 lead-gen campaigns, 3.1x more
5.9% vs 2.6%
Share of clicks that became leads, LinkedIn against Facebook, 2025 lead-gen campaigns, 2.3x more

What should you change?

Change

Rank channels by what a customer cost, not by what a lead cost, and name the numbers you could not work out. LinkedIn’s $202 lead became a $63,312 customer at 0.57x, across 114 advertisers we could follow into the CRM. The other three rows are withheld (why): one ranking without the other is how expensive customers hide inside cheap leads, and an honest blank beats an estimate.

Test

Put both rankings, cost per lead by channel and cost per customer by channel, on the same board slide for two quarters. Where the customer order flips the lead order, the lead price was pricing the wrong event; narrate it in one sentence: “our cheapest lead was our most expensive customer.”

Do not conclude

“Kill Meta.” Facebook delivered 2025 volume at $145 a lead, across 71 advertisers. What a Facebook customer cost is withheld (why), which means this page cannot price it, not that it is expensive. A missing number is a reason to measure the channel, never a reason to cut it.

Change

Wire closed-won revenue back to the channel that sourced it this week, so cost per customer is a standing report instead of a quarterly dig: LinkedIn’s $63,312 at 0.57x stays invisible until that link exists. Tag channel and audience the moment a lead arrives and carry the tag through the CRM to the deal. Then check that no single account is carrying the whole result before you show anyone the table.

Test

Flag every channel that moves two or more places between the two rankings, cheapest lead against cheapest customer. Run the budget conversation on those flags; the gap between the two lists is where the next shift lives.

Do not conclude

“The platform’s own return number is close enough.” Ad platforms re-create the exact illusion this page documents. Only following each lead into the CRM produced the one customer number here, $63,312 at 0.57x, and only that same link tells you when one account is carrying so much of the result that the number means nothing.

What are the exact numbers?

All four channels, both questions, against the whole dataset. What a lead cost comes from 2025 lead-generation campaigns; what came back and what a customer cost come from the advertisers we could follow into the CRM, counting only revenue we can trace back to an ad. Where a number is marked withheld it exists in our working set and we chose not to publish it, because it would describe one advertiser rather than a market.

ChannelAdvertisers (leads · followed to revenue)Cost per leadBack per ad dollarCost per customer
LinkedIn138 · 97$2020.57x$63,312
Instagram49 · 34$138withheldwithheld
Facebook71 · 49$145withheldwithheld
Google Ads56 ·, $524withheldwithheld
Every advertiser we could follow into the CRM (all channels)1270.56x$58,887

The withheld numbers: on Facebook and on Instagram one account held more than 40% of the closed-won revenue, so neither number describes a market, and Google Ads had too few advertisers to publish. All four are listed with their reasons in the list of what we killed. They are drawn hatched in the chart above, never estimated. Lead prices come from $31.5M of lead-generation spend; what came back and what a customer cost come from the $29.4M, 154k leads and 127 advertisers we could follow into the CRM, out of $57.6M of analysed 2025 spend.

How do we know?

Every number above had to clear the same rules before it reached this page. In brief:

  • How many advertisers are behind each number: 138 for the LinkedIn lead price, 71 Facebook, 49 Instagram, 56 Google Ads; 97 behind the LinkedIn customer number we published. The count is printed beside every number on this page.
  • No one advertiser is allowed to swing a number: in any group we publish, no single advertiser may hold more than 40% of the spend or more than 40% of the closed-won deals. The deals half is what removed the Facebook and Instagram customer numbers on 2026-08-17: their leads were fine, their wins belonged to one account. Google Ads had too few advertisers to follow into revenue. Withheld, never estimated, reasons in the list of what we killed.
  • Spend floor: at least $50,000 spent in any group we publish.
  • Scope and window: lead prices, click prices and conversion come from $31.5M of lead-generation spend inside the $57.6M we analysed for 2025. What came back and what a customer cost come from the $29.4M, 154,000 leads and 127 advertisers we could follow through the CRM to closed-won revenue over the following 12 months. We counted revenue only where the ad-generated lead is the one that opened the deal. This is credit we can trace, not proof the ads caused the sale, and it is not profit and not lifetime value.
  • We watched, we did not experiment: every figure is something that happened in 2025, not proof that one thing caused another, it describes these advertisers, not a guaranteed result for you.
  • What else could explain this: the job each channel was given. Plenty of advertisers run Facebook for volume rather than pipeline, and a channel bought for reach will not look like a closer. We cannot test that here, because the number that would show it is withheld. On the published side, LinkedIn’s 0.57x sits close enough to the 0.56x everything else managed that the channel you pick explains very little of the shortfall. What each campaign was bought to do, which accounts are in the mix and which industries they sell into are all uncontrolled here.

What do you do different tomorrow morning?

One named change per seat. The columns reorder to match your reading lens; both are always on the page.

Share this finding

Send the playbook to your demand-gen lead

Send the business case to your CMO

Three lines, the claim, the decision, the source, ready to paste.

With your team Email your demand-gen leadEmail your CMO
With the world
If you’re the CMO

Change the board slide

The one change: put both rankings side by side, cost per lead by channel next to cost per customer by channel, on the same slide, every reporting cycle, starting this one.

  1. Pull both rankings for the last two quarters and put them next to this page’s table. If your channels change places the way these did, the slide explains itself.
  2. Narrate the inversion in one sentence: “our cheapest lead was our most expensive customer.” That sentence, not a CPL trend line, is what reprices the mix.
  3. Judge every channel move on pipeline per dollar we can trace back to an ad; a report that only shows the lead price will re-argue for the cheap-lead channel every quarter.
If you’re the demand-gen lead

Stand up cost per customer, by channel

The one change: wire closed-won revenue back to the channel that sourced it this week, so cost per customer by channel is a standing report, not a quarterly dig.

  1. Tag every campaign’s leads with channel and audience the moment they arrive, and carry the tag through the CRM to the deal. That link is where most of these reports die.
  2. Build the two-quarter look-back: pipeline and closed-won revenue by channel, traceable back to an ad, next to what you spent. Re-rank every quarter.
  3. Flag any channel that moves two or more places between the two rankings. That gap is where the next budget conversation lives.

Your channel ranking is only as good as the number you ranked it on.

Bring what a lead and a customer cost you by channel for the last two quarters and we’ll put them next to the benchmark, and show you which channels earn their place and which just look cheap.

Benchmark my channels