Creating demand cost about what chasing it did: $187 vs $196 a lead.
There is no cheap half of the budget.
Buying attention from people who have never heard of you cost what chasing the people who already have did: $187 a lead against $196 in metadata.io’s 2025 B2B benchmark, 105 advertisers and $21.42M of spend on one, 85 advertisers and $11.41M on the other. So neither half of the budget earns a built-in discount next quarter, and the argument you can actually win moves one level down: on the same cold audiences, the ad format you pick went with a 2.4x difference in what a lead cost. Two honest limits, and they belong here rather than at the top: nine dollars apart is close enough to call a tie, and the two audience numbers come from two different sets of advertisers doing two different jobs, a starting point, not a ranking.
Change the format before you change the audience split. On cold prospecting audiences in metadata.io’s 2025 benchmark, a document ad made a lead for $148 at 10.6% click-to-lead against $358 for video, and that cheap format took only 8.5% of cold spend. On cold audiences Meta bought the cheaper lead and LinkedIn turned more clicks into leads, so run both for a fortnight before anyone reopens the cold-versus-retargeting argument: those two lines came in at $187 and $196, close enough to call a tie, and on two different sets of advertisers at that.
2025 · 105 advertisers on cold prospecting, 85 on retargeting · lead-generation campaigns · only groups with at least $50K spent and 5 or more advertisers · Methodology · Suppression rules · Typical vs best-case: how to read this
Read this before you move a budget
This is not a reason to cut retargeting. First-touch cost per lead is not retargeting’s job; its pools are smaller, capped and downstream of cold spend.
The $187 and $196 lines are close enough to call a tie, and they are not apples to apples: 105 advertisers sit behind one and 85 behind the other, running different channels, formats and offers. Treat cold-against-retargeting as something to test on your own account, and take the format and channel cuts below as the part you can put in a plan.
Executive summary
- The finding you can plan on: format, not whether the audience is cold or warm. On cold audiences, document ads made a lead for $148 against $177 for image and $358 for video, a 2.4x spread inside one apples-to-apples cut, where the cheapest format takes 8.5% of cold spend.
- The tie, stated plainly: creating demand cost $187 a lead and chasing it $196. That is close enough to call a tie, not a premium and not a ranking, two different sets of advertisers, different in size and mix (105 against 85), doing two different jobs. The budget call it supports is that neither line deserves a built-in efficiency discount in the plan.
- The limit, stated plainly: this page prices leads, not revenue. What retargeting leads became is withheld (why), one advertiser held too much of that result to publish it, so nothing here says a cold lead is worth more in the end, and nothing here is a case for cutting retargeting.
Playbook
- The play: change the format before you change the audience split. On cold audiences documents made leads at $148 at 10.6% click-to-lead against $358 for video, and only 37 of 105 cold advertisers run one.
- The split: create on Meta ($153 Instagram, $166 Facebook), convert on LinkedIn (6.6% click-to-lead against 3.9% and 2.5%). Two jobs, two budgets, one offer.
- What this is not: a retargeting cut. Cold at $187 and retargeting at $196 is close enough to call a tie on two different sets of advertisers, and your retargeting pool exists because cold spend filled it. Keep it funded, test the format.
What actually changes the cost of a cold lead?
The format, by a wide margin, and the cheapest one is the one almost nobody runs. On cold prospecting audiences, document ads made leads at $148 with a 10.6% click-to-lead rate (37 advertisers), against $177 for image at 4.7% (99), $325 for conversation ads (26), $358 for video at 2.5% (48) and $617 for carousel (19).
This is the cut that survives the hardest reading of the data, and it is why it leads the page. Every row here is the same audience type, the same year and the same publication rules, apples to apples, unlike the cold-against-retargeting question further down. The spread between the cheapest and the most expensive format is 2.4x in cost per lead, and the direction runs against how plans are written: the format that creates demand most cheaply takes 8.5% of cold-audience spend, and only 37 of the 105 advertisers running cold campaigns use it at all, while video and carousel, the two most expensive rows. Are the two most plans lead with.
Where should the cold budget sit, Meta or LinkedIn?
Both, with different jobs. On cold audiences Instagram bought the cheapest lead at $153 (37 advertisers) and Facebook at $166 (59), against $194 on LinkedIn (94). But LinkedIn turned 6.6% of its cold clicks into leads, against 3.9% on Instagram and 2.5% on Facebook. Meta is where cold attention is cheap; LinkedIn is where a cold click is most likely to leave a name. The honest play is to run both and hold them to different numbers, not to crown one demand-creation channel.
So is creating demand cheaper than chasing it?
Not in any way you should plan against. On 2025 B2B lead-generation campaigns, audiences built to reach people who had never touched the brand produced 114,581 leads from $21.42M, a $187 cost per lead across 105 advertisers. Retargeting, the same offers shown back to site visitors and known contacts, produced 58,087 leads from $11.41M across 85 advertisers, at $196.
Nine dollars apart is a tie, not a finding. It is also not apples to apples: different advertisers, different channels and formats, different offers, and two groups of different sizes. What the two numbers do kill is the discount. The assumption that chasing demand you already created is the structurally cheap half of a budget and creating it is the expensive bet. Across these advertisers there was no such bargain, so neither line should carry a built-in discount into next quarter’s plan. What they do not license is a cut: first-touch cost per lead is not what a retargeting pool is for, and that pool exists because cold spend filled it.
What should you change?
Fund a format line in the plan before you touch the audience split: document creative on cold audiences made a lead for $148 against $358 for video, on one apples-to-apples cut (n=37 and 48). It is the biggest fair comparison on this page, and the only one that does not require re-arguing the audience mix.
Approve a budget-neutral format swap, one video-led cold campaign rebuilt as a document ad, same audience, same offer, same quarter, and ask for three numbers back: cost per lead, click-to-lead, and CRM closed-won per dollar. The first two land in weeks; the third decides next year’s split, so set the window to at least twice your sales cycle before the test starts.
“Cold is cheaper, so cut retargeting.” $187 against $196 is close enough to call a tie on two different sets of advertisers, and this page prices leads rather than revenue, what retargeting leads became is withheld, because one advertiser held too much of that result. First-touch cost per lead is not retargeting’s job; its pool is smaller, capped and downstream of the cold spend that filled it.
Launch one cold prospecting campaign this week as a document ad, ungated, 8–12 slides, mid-funnel: $148 per lead at 10.6% click-to-lead on cold audiences (n=37 advertisers). You are not adding a channel or an offer, only a format and an audience your plan already funds elsewhere.
Split the cold budget by job: Meta for reach at $153–$166 per lead, LinkedIn for conversion at 6.6% click-to-lead. Run the same offer on both for four weeks and report them separately. A blended cold CPL will bury exactly the difference you are testing for.
“Cold is the cheap half of the budget, so move the money.” The two lines are $187 and $196, close enough to call a tie, on different sets of advertisers. $187 is also an average across 105 advertisers, weighted by how much each of them spent, not a promise about your next campaign, and a cold audience broadens as you spend into it. Scale in increments and watch cost per lead and click-to-lead per increment.
What are the exact numbers?
Three cuts of the same question, all about what a lead cost: 2025 lead-generation campaigns, averaged with each advertiser weighted by what it spent, and published only where a group held at least 5 advertisers, $50,000 of spend and no single advertiser above half of it. The two apples-to-apples cuts, format and channel inside cold audiences, come first; cold against retargeting comes last because it is the weakest of the three. Nothing on this page follows a lead into revenue.
Ad format, cold prospecting audiences only
| Ad format on cold audiences | Advertisers | Cost per lead | Click-to-lead | Share of cold spend |
|---|---|---|---|---|
| Document | 37 | $148 | 10.6% | 8.5% |
| Image | 99 | $177 | 4.7% | — |
| Conversation | 26 | $325 | — | — |
| Video | 48 | $358 | 2.5% | — |
| Carousel | 19 | $617 | 1.5% | — |
Dashes: numbers that did not clear the advertiser or spend minimum on their own, or that are published in the interactive explorer rather than quoted here. Share of cold spend is printed for document ads because how few advertisers run them is the finding; the remaining formats split the rest. Every row is the same audience type, the same year and the same publication rules. Which is what makes this the apples-to-apples cut on the page.
Channel, cold prospecting audiences only
| Channel on cold audiences | Advertisers | Cost per lead | Click-to-lead | The job it is good at |
|---|---|---|---|---|
| 37 | $153 | 3.9% | Cheapest cold lead | |
| 59 | $166 | 2.5% | Cheap reach at volume | |
| 94 | $194 | 6.6% | Converts a cold click best |
Advertiser counts overlap: most of these advertisers run cold audiences on more than one channel, so the three rows do not add up to the 105. We could not publish a cold-audience number for Google Ads under the same rules, so it is left out rather than estimated.
Creating demand vs chasing it. The cut that is not apples to apples
| Audience type | Advertisers | Spend | Leads | Cost per lead | Click-to-lead |
|---|---|---|---|---|---|
| Creating demand (cold prospecting) | 105 | $21.42M | 114,581 | $187 | 5.1% |
| Capturing demand (retargeting) | 85 | $11.41M | 58,087 | $196 | 4% |
No row in this table is highlighted, because the two are not apples to apples and the difference between them is too small to call a winner: different advertisers, different channels, formats and offers, and two groups of different sizes (105 against 85). What each lead was worth is deliberately absent as well: what retargeting leads became is withheld (why) because one advertiser held too much of that result, so we cannot publish a revenue comparison between these two rows.
How do we know?
Every number above had to clear the same rules before it reached this page. In brief:
- Advertiser counts: 105 advertisers ran cold prospecting audiences and 85 ran retargeting inside 2025 lead-generation campaigns; format and channel rows rest on 19–99 advertisers each, and n is printed beside every one.
- Why cold against retargeting is a question, not a finding: it is not apples to apples. The two sides hold different advertisers in different numbers, running different channels, formats and offers, and the gap between them is small enough to sit inside that variation. Retargeting pools are also small, capped, quickly worn out and downstream of cold spend, so pricing a first touch against them prices something retargeting is not run to do. We publish it because it kills the assumed discount, and we refuse to publish it as a reason to cut retargeting.
- What has to be true before we print a number: at least 5 advertisers in the group, $50,000 of combined spend, and no single advertiser above 50% of that spend. Groups that miss any of those are withheld with the reason given, never estimated or folded into a bigger number.
- Scope and window: calendar-2025 lead-generation campaigns from the $57.6M analysed. Cost per lead is spend ÷ leads for each group, with each advertiser weighted by how much it spent; click-to-lead is leads ÷ clicks. “Creating demand” is cold prospecting targeting, firmographic, technographic, intent and account-list audiences of people with no prior touch. “Capturing demand” is retargeting.
- Nothing on this page follows a lead into revenue: we could not publish what retargeting leads paid back or what a retargeted customer cost, one advertiser held too much of that result. That is why this page compares what a lead cost and how often a click became one, and stops there.
- We watched, we did not experiment: these are things that happened together in 2025, not proof that one caused the other. They describe what these advertisers did, not what is guaranteed to work.
- What else could explain this: retargeting pools are small and finite, so an advertiser who spends into one shows the same people the same ad more often and pushes cost up with it, part of the $196 may be pool exhaustion rather than audience quality. Offer mix is the other candidate: cold campaigns lean on mid-funnel content while retargeting leans on demo requests, which convert differently regardless of temperature. And a retargeting lead is, by construction, someone the brand already paid to reach. The cost of creating that first touch sits in the cold line, not the retargeting one. We cannot rule any of those out here, which is why cold against retargeting is published as a tie and a question rather than as a ranking. The format cut carries none of those problems.
Full rules in the Closed-Won Protocol, including the claims we refused to publish, with the exact reason each one was killed and the number it cost us. New here? Start with typical vs best-case: how to read any benchmark.
Cite as: Metadata 2026 B2B Ad Spend Benchmark (n=153 advertisers, $57.6M, 2025), metadata.io/benchmark-report-2026
What do you do different tomorrow morning?
One named change per seat, both executable inside two weeks. The columns reorder to match your reading lens; both are always on the page.
Buy the format gap, not the audience argument
The one change: fund a document-format test on the cold audiences you already run, and stop pricing either audience type at a discount.
- Ask for the split before the plan: what percentage of last quarter’s paid budget created demand, what percentage harvested it, and what formats carried each? Most teams cannot answer that in one number, which is itself the finding.
- Price the two audience lines honestly: $187 per lead against $196 is close enough to call a tie on different sets of advertisers. Retargeting keeps its job and its budget; what it loses is the assumption that it is the cheap half.
- Move the argument to the apples-to-apples cut: $148 against $358 per lead by format on the same cold audiences. That is where a decision this quarter has evidence behind it.
- Set the verdict metric before the money moves: CRM closed-won per dollar by format and audience type, over at least twice your sales cycle. Cost per lead alone is how a tactic gets funded on the wrong scoreboard.
- Read your own allocation against the market in finding 02 before you set next quarter’s split.
Two weeks, two changes
The one change: launch a cold prospecting campaign as a document ad, and split the cold budget by job across Meta and LinkedIn.
- Week one: take an asset you already have, a teardown, playbook, benchmark deck, cut it to 8–12 slides, put the payoff in the first three, and run it ungated to a cold firmographic or intent audience.
- Week one, same day: keep your current image ad live as the control on the same audience and offer, so the readout is clean.
- Week two: duplicate the cold campaign across Meta and LinkedIn with the same offer, and report the two separately, Meta is buying reach at $153–$166 per lead, LinkedIn is buying conversion at 6.6%.
- Report cost per lead and click-to-lead per increment, then CRM closed-won per dollar when the cycle allows, and never a blended cold number, which hides both effects at once.