Benchmark methodology.

The five rules we don't break

• No blended numbers. Third-party platform benchmarks and Clever Zebo's own client-account data are stored and shown as separate tiers — never averaged into one hybrid figure.
• No fabricated data. If there's no real sourced data point for an industry × channel, you get an explicit "insufficient data" message — never an interpolated or estimated guess.
• No silent staleness. Every figure carries a source name, sample, report period and last-verified date, shown inline as "as of [date]."
• CPL isn't the finish line. The tool always pushes past cost per lead toward cost per customer (CPL ÷ close rate), with the formula shown.
• A median is shown as a median. Our current sources publish a single median figure, so we display a median. We never dress one number up as a range it never was.

Where the numbers come from

We only use named, dated sources, and we cross-check every figure against the original report before it goes in. We reject content-farm posts that recite "2026 benchmarks" with no methodology and no date. We don't link out to the sources (we'd rather not hand competitors our backlinks), but we name every one, show its sample, and flag when a source doesn't disclose its sample size so you can weight it yourself.

Two lenses we never average together

"Cost per lead" means different things in different studies, so we keep two lenses strictly separate:

• Platform conversion CPL — what a single conversion (a form fill, call or lead-form submission) costs on one ad platform. This is the WordStream/LocaliQ data for Google Search and Meta, and The B2B House data for LinkedIn.
• Qualified-B2B-lead CPL — what a sales-ready B2B lead costs across all your paid or organic channels. This is First Page Sage's data, shown under the "Paid (all channels)" and "Organic / SEO" channels.

These answer different questions, so their numbers differ — often a lot. A legal lead reads as $131.63 on Google Search (a platform conversion) but $784 as a cross-channel qualified lead. Both are real; neither is wrong; we never average them into a single figure. The calculator and every page label which lens a number belongs to.

Platform benchmarks vs. Clever Zebo client data

Today, every figure in the tool is a platform benchmark — an industry-wide number from a third-party research report. Separately, Clever Zebo manages real client ad accounts, and that first-party data is often more relevant than any industry average. When we publish it, it will appear as its own clearly-labelled tier ("from Clever Zebo's own accounts, n=X"), with the account count always shown, and it will never be mathematically combined with platform benchmarks. We also won't show a client-data figure at all unless it's backed by a minimum sample of at least three distinct client accounts, so a single outlier account can't masquerade as a trend.

How we map categories

Different sources slice the world into different industry buckets, so we map each source's category names onto one canonical list — and we log every mapping decision rather than silently assuming, say, that "Business Services" means "SaaS." Where a mapping is loose, we lower the confidence rating on that data point and disclose the caveat right on the result. Here are the mapping decisions behind the current data:

Confidence ratings

Each data point is rated high, medium or low, reflecting two things: how cleanly the source's category maps to the industry you picked, and how transparent and large the underlying sample is. WordStream/LocaliQ figures (a disclosed 13,000+ campaign sample) earn high on a clean mapping. First Page Sage's B2B cross-channel figures are medium — named and methodical, but the sample size isn't disclosed. The single-aggregate LinkedIn data is low. The rating shows on every result so you can weight it accordingly.

Known limitations (v1)

• US only. Every figure is United-States data. We don't yet publish geographic segmentation.
• LinkedIn is a single aggregate. The only LinkedIn-by-industry CPL data with any stated methodology comes from one agency aggregate (~$1M spend over six months) with an undisclosed account count. We include it, rated low-confidence and clearly flagged — not as gospel.
• Undisclosed samples. First Page Sage and The B2B House describe their method but don't publish a sample size (N). We label these directional rather than treating them like population studies.
• No Google Display / PMax. No source we trust publishes a verifiable per-industry Display CPL, so we don't show one.
• No company-size split. LinkedIn costs in particular vary several-fold by seniority and company-size targeting; we don't fake that granularity.
• Curated, not live. This is manually-sourced, versioned data, refreshed on a cadence — not a scraped or auto-updating feed.

Refresh cadence

We refresh data whenever our sources publish updated reports (historically once or twice a year), and at each refresh we re-verify that every figure still matches the original source — reports do get revised. Any Clever Zebo client-data tier is reviewed quarterly, re-checking the minimum-sample threshold each time.