What $57.6M of B2B ad spend says you should do differently.
153 advertisers · $57.6M · 211,000 leads · 2025 · no email gate. You own the budget, so every finding here is an allocation decision, measured against what the market actually did.You run demand generation, so every finding here is a play, with the format, channel and audience named.
Two questions, two doors
This report answers exactly two questions, and which one you have depends on what you own. Pick your door. The whole report re-ranks itself behind it. Prefer a straight reference table? B2B advertising benchmarks by channel lists CTR, CPC, CPM and cost per lead with n on every row.
What should I do with paid to create demand?
Creating demand cost $187 per lead against $196 for harvesting it, close enough to call a tie, not the premium most plans budget for. The gap worth acting on is ad format: inside cold audiences the cheapest lead came from a format almost nobody runs ($148 against $358 for video), and the cheapest channel is not the best-converting one. Every play here is executable in two weeks.
Start with the demand-creation playbook →How do I measure against the market?
The typical B2B advertiser put 25% of budget into creating demand, 7% into retargeting and 16% into buying clicks. We publish the whole range, the quietest quarter, the middle and the boldest quarter, so you can see where you sit and defend it, instead of arguing from a recommended mix nobody ever measured.
Open the allocation benchmark → Then price your own numbers in the Reality Gap calculator, your CPL placed in the real per-advertiser range, no email gate, prints as a board page.Everything below is the evidence for those two answers: ten findings, each built from enough advertisers to publish, checked so that no single advertiser can swing a number, and recomputed from raw totals on every build. Where a number failed those rules we withheld it and said why, rather than estimating it. If you read benchmarks for a living, start with typical vs best-case: how to read any benchmark, the five questions we put our own numbers through.
If you only do one thing: the Monday list
Seven moves, taken straight from the ten findings below. Each names the change, the number behind it and the page that proves it. Five of them are a settings change or an afternoon of work; two are a planning decision.
Switch your LinkedIn lead campaigns to the built-in forms.
Same offer, one setting: $193 a lead through the in-feed form against $346 when you send people to a landing page. Keep landing pages for people who already know who you are.
Rebuild your best-performing offer as a document ad.
Inside cold audiences a document ad made a lead for $148 against $358 for video, and almost none of your competitors are running the format.
Cap your click campaigns and take them out of your cost-per-lead reporting.
Across the dataset, $12.7M bought clicks and produced 66 leads. Not proof the money vanished. Proof that lead reporting cannot see it. Give each one a stated brand purpose and a number it has to hit, or stop funding it.
Find out where your own cost per lead actually sits.
Three out of four advertisers paid more than the $202 everyone quotes. If you are above it you are probably normal, not behind, and that is the argument to bring to your next budget conversation.
Split your campaigns by company size before you scale them.
A customer cost $35,288 targeting 51–200-employee companies on LinkedIn and $130,468 at 501–1,000. One blended average hides that completely.
Judge audiences on customers, not on cost per lead.
The pricier audience produced 3.56 customers per 1,000 leads against 2.6. A cost-per-lead dashboard will keep defunding your best audience.
Before you plan next quarter, write down your own three percentages.
How much you spend creating demand, retargeting and buying clicks. The typical advertiser sat at 25% / 7% / 16%. An afternoon of work that settles a year of arguments.
None of these is a promise about your account. They are the moves this dataset supports, with the number attached to each one so you can argue with it. What the data cannot tell you, whether the revenue would have arrived anyway, is listed in full under the questions this data cannot answer.
Typical vs best-case: how to read any benchmark
Two very different numbers circulate in this market, and they get quoted as though they were the same thing. A benchmark tells you what to expect, the number you plan against. A case study tells you what is possible, one customer’s best year, published because it went well. Five questions tell you which one you are holding, and they work on our numbers as well as anyone else’s.
- 1. Who is in it Was the group picked before anyone saw the results, or picked afterwards because the results looked good?
- 2. Divided by what A return on how much spend, over how long? A number with no spend behind it can be made to say anything.
- 3. What is actually counted Revenue traced back to an ad, or pipeline the campaign was merely near? Closed deals, open pipeline and MQLs are three different things and do not compare.
- 4. How many, and how spread out How many advertisers is it built on, and how wide is the range? One number with neither is a story, not a benchmark.
- 5. Can anyone check it Could someone else rebuild the number from what is published, and does the publisher correct itself in public when it turns out to be wrong?
How a vendor’s biggest case study and this report’s payback floor coexist
“122X is one customer’s influenced-pipeline number, published by a vendor because it went well. 0.56x is what a traced ad dollar returned in closed-won revenue across 127 advertisers. Different measure, different group, different job, they only contradict each other if you use one customer’s best year as your forecast.”
The rule this report holds itself to: a benchmark tells you what to plan for; a single success story only proves a much better result is possible. The two should never appear on the same axis, not in a chart, not in a table, and not in a sentence that averages them into a range.
Ten findings from $57.6M of B2B ad spend
The first two answer the two questions above; the rest are the evidence underneath them. Each one is a claim we can defend: built from enough advertisers to publish, checked so no single advertiser can swing it, and rebuilt from the raw totals every time the page is built. Click through for the proof. The chart, the numbers, how we got them, and what to do about it.
Creating demand cost about what chasing it did: $187 vs $196 a lead.
The typical B2B advertiser spends 25% of budget creating demand, and 16% buying clicks.
$12.7M went to click campaigns and 99.4% of them recorded no lead at all.
LinkedIn document ads made a lead for $142. Most advertisers skip the format.
LinkedIn's built-in forms cost $193 a lead. Landing pages cost $346.
The $202 everyone quotes? Three out of four advertisers paid more.
Instagram bought the cheapest B2B lead we found: $138. Worth a test.
Chasing mid-market cost $130,468 a customer. Small companies on LinkedIn cost $35,288.
The pricier leads won more customers: 3.56 per 1,000 against 2.6.
Ads paid back 0.56x in year one, 56 cents on the dollar.
Find companies like yours
Narrow $57.6M of spend down to the advertisers who look like you, your company size, your industry, the channels and ad types you actually run, and see what they paid and what they got. Every view has its own link you can send to someone. Where too few advertisers stood behind a combination to publish it safely, we grey it out before you click rather than showing you a number we do not trust.
Ad channel, ranked
Lead metrics (CPL, conversion rate) are computed across lead-generation campaigns; delivery metrics (CPM, CPC, CTR) across all campaigns in the segment. Pipeline metrics cover the $29.4M lead-gen subset with CRM attribution (the subset is $31.5M before the outcome-concentration exclusion). Full methodology.
Take the whole report with you
No forms, no gate. Share it, cite it, forward it to whoever owns the budget.
The full report (PDF)
Key findings, both deep-dives and the methodology in one document.
Download the PDFBuilt for sharing: the CMO edition · the demand-gen edition
The data itself (CSV)
Every published cut as a spreadsheet, company size, industry, channel, ad type, campaign goal, offer and audience, with the advertiser count behind each row. Aggregated and anonymized; suppressed cuts are absent rather than zeroed.
Download the CSVMachine-readable: the same cube as JSON · CC BY 4.0
You own the budget?
Start with the allocation benchmark: where the market puts its money across creation, retargeting and traffic objectives, printed as a range you can find your own split in, then check your own cost per lead in the calculator.
Start with finding 02You run demand generation?
Start with what to do with paid to create demand: cold and retargeting priced close enough to call a tie, document ads made the cheapest cold lead by a wide margin, and Meta and LinkedIn do different jobs on the same cold audience.
Start with finding 01How this data earns trust
The dataset. 153 B2B advertisers and $57.6M of calendar-2025 spend across LinkedIn, Google Ads, Facebook and Instagram; 153 advertisers appear in at least one published segment. Reddit and Microsoft Ads spend is in the totals but too thin to publish as segments. Pipeline metrics cover the $29.4M subset of lead-generation campaigns with CRM opportunity attribution, 127 advertisers and 154,000 CRM-joined leads.
The thresholds. Media metrics (CPL, CPC, CVR) publish only from segments with at least 5 advertisers, $50,000 of combined spend, and no single advertiser above 50% of segment spend. Pipeline metrics (CAC, traceable payback) are held to a stricter bar: at least 8 advertisers, at least 3 closed-won opportunities, and no single advertiser above 40% of that group’s wins, a rule written down before we saw the numbers. Groups that fail are withheld with an explicit reason, never estimated, never quietly blended into a parent.
The kill list. Eight claims are on it, and it is a feature of this report rather than a footnote. It includes what would have been the strongest headline here, the retargeting pipeline row that failed the outcome-concentration gate, and a close rate that turned out to be computed on an export with no closed-lost opportunities, re-published as a true won-vs-lost rate after the re-export, pointing the other way. Every cut is listed with the rule it broke and the number it cost us.
What happened downstream. On the $29.4M subset that carries CRM attribution, honestly computed payback came to 0.56x at a $58,887 CAC, attributed, not incremental, and published as a starting point rather than a verdict. It sits at finding 10, next to what we refused to publish.
Read the full methodology & data standards · See what we refused to publish and why
Cite this data. Metadata 2026 B2B Ad Spend Benchmark (n=153 advertisers, $57.6M, 2025). Https://metadata.io/benchmark-report-2026, the aggregate JSON is available under CC BY 4.0.
Want your numbers benchmarked against this dataset?
Metadata runs paid campaigns for 150+ B2B companies. Book a demo and we'll show you where your CPL, CAC and pipeline ROI sit against these benchmarks, and what to change.
Book a Demo