B2B ABM & paid media benchmarks 2026

The short version

B2B paid media in 2026 costs roughly $130,000 to acquire a mid-market customer and about $35,000 to acquire a small-company customer, according to the largest public dataset available, Metadata.io's 2026 B2B Benchmark Report, which analyzed $57.6 million of B2B ad spend and tied it to closed-won revenue rather than leads (source 1, 2). A lead from a demand-creation campaign costs about the same as a lead from a demand-capture campaign, $187 versus $196, yet most B2B budgets put only about a quarter of spend into creating demand (3, 4). And a large share of B2B ad money still goes to click-and-traffic objectives that produce almost nothing: $12.7 million of the analyzed spend went to traffic campaigns, and 99.4% of it recorded no lead at all (5).

This page is the hub. It pulls together the public benchmarks that matter for account-based marketing and demand generation, puts them in tables, and tells you where each number comes from so you can decide how much to trust it. Every figure links to a public source. Where the sources disagree, we say so. Where a number is a self-reported survey average rather than measured spend data, we say that too, because it changes what the number is worth.

How to read a B2B benchmark before you trust it

A benchmark is only as good as the population and the metric behind it, and most B2B benchmarks are weaker on both than their headlines suggest. Before you put any number on this page into a board deck, check three things.

1. Is it measured or surveyed?

Measured benchmarks come from ad-platform or CRM data: actual spend, actual clicks, actual opportunities. Metadata's 2026 report is measured, built on $57.6 million of spend run through its platform and matched to closed-won deals under what the company calls its Closed-Won Protocol (6). Surveyed benchmarks come from asking marketers what they think their numbers are. Demandbase's 2024 ABM Benchmark and the Demand Gen Report benchmark surveys are surveys (7, 9). Survey data is useful for adoption, maturity and budget questions. It is unreliable for cost metrics, because respondents round, forget, and flatter themselves.

2. What is the unit?

A "lead" on LinkedIn means a Lead Gen Form submission. A "lead" in a CRM report might mean an MQL, a demo request, or anyone who downloaded a PDF. Metadata's cost-per-lead insight explicitly warns about this: cheap leads from broad campaigns often produce zero pipeline, which the company calls the CPL illusion (8). When you compare a CPL from one source with a CPL from another, make sure both mean the same thing. Often they do not.

3. Who is in the sample?

Metadata's customers skew toward venture-backed B2B software companies running paid social and paid search. Demandbase's survey respondents skew toward companies that already bought an ABM platform. 6sense's data reflects accounts inside its own platform. None of these is the whole B2B market. Each is a large, useful, biased slice, and the bias is different for each. We note the slice next to every table on this site.

For a longer walkthrough of this, read the guide to reading benchmarks and the site methodology page.

Customer acquisition cost benchmarks by company size

The single most important B2B paid media benchmark is what it costs to acquire a customer, and the best public data shows it climbs steeply with company size. Metadata's 2026 report puts the paid-media cost per customer at $130,468 for mid-market companies and about $35,000 for small companies, a gap the company calls the CAC cliff (2). The mid-market number is not a typo. It reflects longer sales cycles, larger buying groups, and much lower conversion rates from lead to closed-won as deal size grows.

Two cautions. First, these figures are paid-media CAC, meaning ad spend divided by closed-won customers attributable to that spend. They exclude headcount, tools and agency fees, so a fully loaded CAC will be higher. Second, the number is an average across many companies and product categories. A $130,000 cost per customer is fine for a $60,000 ACV product with strong retention and disastrous for a $12,000 ACV product. Read the B2B CAC benchmarks page for how to convert this into a payback target.

The practical lesson from the CAC cliff is that the mid-market is not "SMB with bigger logos." Companies that move upmarket on the same paid-media playbook they used for SMB routinely see cost per customer triple or quadruple, which is roughly the ratio the Metadata data shows.

Cost per lead benchmarks by channel

Across B2B paid channels, cost per lead ranges from roughly $30 on the low end of Meta and Google Search up to well over $200 on LinkedIn for enterprise software audiences, with the measured B2B average for paid social sitting close to $190 per lead. Metadata's create-vs-capture analysis, which covers the full dataset, reports $187 per lead from demand-creation campaigns and $196 from demand-capture campaigns (3). The channel-level ranges in the table below come from the sources noted in each row, and they are not all measured the same way.

The most useful thing in that table is the last two rows read together. Demand creation and demand capture leads cost within 5% of each other in measured data, which undercuts the common belief that "top of funnel is cheap and bottom of funnel is expensive." It also means the CPL comparison is not where the decision should be made. The decision is about what those leads turn into, which is the subject of the create versus capture demand page.

The traffic-objective row is the clearest waste signal in any current B2B benchmark. Click and traffic objectives optimize for the cheapest possible click, and the cheapest possible click in B2B comes from people who will never buy. Metadata's data found $12.7 million of analyzed spend going to those objectives with a 99.4% no-lead rate (5). If your agency reports CPC as a primary KPI, this is the number to bring to the next meeting.

CPC and CTR benchmarks by channel

Cost per click in B2B runs from under $1 on Meta to $5 to $10 or more on LinkedIn, and click-through rates run from about 0.4% on LinkedIn Sponsored Content to 1% or higher on Meta and 3% or higher on Google Search brand terms. These are the widely published ranges, and they are the least important numbers on this page. CPC and CTR are input metrics. They tell you whether the ad and audience are working mechanically, not whether the spend produces revenue.

Two observations. The LinkedIn CTR floor of roughly 0.4% is the number most B2B marketers benchmark against, and being above it says almost nothing about whether the campaign is producing pipeline. Metadata's CPL illusion analysis makes the same point about CPL: an above-benchmark CTR feeding a lead form that collects students and job seekers is a worse outcome than a below-benchmark CTR feeding a form that collects directors at target accounts (8). Second, display CTRs in the 0.1% range are normal for account-based advertising and are not a failure; the metric to watch there is account engagement lift, which we cover in the ABM engagement benchmarks.

Budget allocation benchmarks: create versus capture

B2B marketers put about 25% of paid budget into creating demand and the remaining roughly 75% into capturing demand that already exists, according to Metadata's 2026 budget allocation analysis (4). That split matters because the measured cost per lead of the two approaches is nearly identical ($187 versus $196), which means the imbalance is not being driven by cost efficiency (3). It is being driven by attribution: capture campaigns are easier to credit in a last-touch model, so they get the budget.

The full budget-allocation table, with the segment-by-segment breakdown and a discussion of what a rebalanced budget looks like, is on the budget allocation benchmarks page.

ABM program benchmarks: adoption, maturity and buying groups

Program-level ABM benchmarks come mostly from surveys and from analyst research, and they describe how buyers behave and how marketing teams are organized rather than what a click costs. The best-known figures are about the buying group and the buying journey.

These numbers explain the paid-media numbers. If buyers do roughly 70% of their process before talking to sales, then capture-only advertising reaches them late, at the moment when they are already comparing a shortlist. If the buying group is 6 to 11 people, then a "lead" who is one person at the account is a fraction of the actual decision. Both facts push in the direction of account-level measurement rather than lead-level measurement, and both are why ABM programs report engagement and pipeline metrics rather than CPL. See the pipeline conversion benchmarks page for what happens after the lead.

The Demandbase and DGR rows deliberately show "survey distributions" rather than a single headline number. Both reports publish percentages of respondents who say they do X, and those percentages change materially with who answered that year. We summarize each on the reports index rather than pulling one figure out of context here.

Cost per lead benchmarks by industry

Cost per lead varies by industry more than it varies by channel, and the published industry data is almost entirely aggregator data rather than measured closed-won data. The pattern is consistent across sources: financial services, legal, cybersecurity and enterprise software sit at the expensive end, while ecommerce, education and consumer-adjacent B2B sit at the cheap end. The full breakdown, with ranges from each source and notes on sample bias, is on the CPL benchmarks by industry page. A condensed view:

The public benchmark reports, compared

Six public reports supply most of the B2B ABM benchmark numbers in circulation, and they are not interchangeable. The table below is the summary; the reports index has a page for each with key findings and direct links.

A note on the sponsor's report. Metadata.io sponsors this site, and its 2026 report is the most-cited source on this page because it is the only one of the six that publishes measured cost figures tied to closed-won revenue. That is a real methodological advantage, and it would be dishonest to pretend otherwise. It is also a single vendor's customer base, which is a real limitation, and we flag it every time. If another vendor publishes measured, closed-won benchmark data, it will get the same treatment here.

Frequently asked questions