Most benchmark tables are one dataset wearing several hats

Cost per lead benchmarks card showing the number two, the two datasets most figures trace back to
Contents 7 sections

Every agency deck has a slide of cost per lead benchmarks. Almost nobody clicks through to where the figures came from. I spent an afternoon doing that, and what turned up was worse than vague.

Two things stood out. Several of the most-quoted sources publish no sample size and no date range at all. And one widely-shared table contains a figure that contradicts the other numbers on its own row.

This is for you if you have ever pasted an industry average into a budget conversation.

What good cost per lead benchmarks look like

There is a simple test, and most published tables fail it. Does the publisher tell you how many campaigns the figures came from, and over what period?

WordStream’s 2026 study passes. It states 13,474 US search campaigns, running April 2025 to March 2026, producing an all-industry average of $66.69 and an 8.18% conversion rate. You can judge whether that sample resembles your business, because you know what it is.

That transparency should be unremarkable. It is not. Once you start checking, the tables that disclose their working are the minority.

The one that disagrees with itself

LocaliQ’s Facebook benchmarks are the most-shared social figures in this space, and there is a problem sitting in plain sight.

The page lists attorneys and legal services at $18.17 per lead. It lists arts and entertainment at $18.17 as well. Identical to the cent, for two industries with nothing in common.

Now take the other numbers on the legal row. Cost per click of $4.10, conversion rate of 10.53%. Divide $4.10 by 0.1053 and you get $38.94. The row’s own figures imply a lead cost more than twice what the row reports.

One of those numbers is wrong. I do not know which, and neither does anyone quoting it. The page discloses no sample size and no data period anywhere, unlike the same publisher’s Google report, which discloses both.

Do not cite $18.17 as a legal Facebook lead cost. It is the single most checkable error in this category.

When two sources are secretly one

Here is the pattern that makes cost per lead benchmarks look better corroborated than they are.

First Page Sage publishes cost per lead by industry: Financial Services $653, Legal $649, Ecommerce $91. No sample size is given. The methodology note says the report “was prepared by our marketing research team, based on data collected between January 2022 through June 2025”, so a three and a half year window is being presented as a 2026 report.

HubSpot publishes cost per lead by industry too: Financial Services $653, Legal $649, Ecommerce $91. Identical, to the dollar, across all three.

To HubSpot’s credit, it says so. Its own text states the data “combines insights from FirstPageSage’s 2026 report”. Nothing is being hidden.

The problem is downstream. Someone quotes HubSpot, someone else quotes First Page Sage, and a room full of people believes two independent sources agree. They do not. It is one unaudited dataset appearing twice, and the second appearance is borrowing HubSpot’s credibility without HubSpot claiming it.

Four questions to ask before trusting cost per lead benchmarks

Run these before you let a figure into a budget conversation.

How many campaigns or accounts is this from? No answer means no weight.

What period does the data cover? A 2026 headline over 2022 data is a different claim from what it appears to be.

What counted as a lead? If they counted newsletter signups and you count booked calls, the comparison is meaningless before you start.

Who benefits from this answer? A company selling SEO publishing figures where organic beats paid in every industry is not necessarily wrong, but it deserves a second look.

How old are cost per lead benchmarks, really

Older than the headline date, almost always.

First Page Sage’s report carries a 2026 label and states its data was collected between January 2022 and June 2025. That is a three and a half year blend. Averaging across that window smooths out exactly the changes you would want to know about, including a pandemic-era distortion at one end and this year’s search price drop at the other.

LocaliQ’s Facebook page was last updated in October 2025 and gives no collection window at all, so you cannot tell whether the underlying data is one year old or four.

By contrast, WordStream states April 2025 to March 2026, twelve clean months. When someone publishes a tight window, it is usually because the number holds up.

Check the date before the figure. A precise-looking number from an undisclosed period is less useful than a rough one you can place in time.

Why bad cost per lead benchmarks change behaviour

A rough guide is harmless. A target changes behaviour, and benchmarks get treated as targets constantly.

I have taken over accounts where a profitable campaign had been paused for running above an industry average, and accounts praised for beating one while losing money on every sale. In both cases the published number outranked the client’s own economics in the decision, which is exactly backwards.

Most cost per lead benchmarks are two datasets wearing twelve different hats. Your margin and close rate are yours alone, and they beat any table.

The honest position on Meta is worth stating plainly too. There is currently no rigorous, sample-disclosed 2026 benchmark for Facebook lead costs. Saying that is more useful than quoting a figure nobody can trace.

What to change this week

Three things.

Open the source behind whichever benchmark you currently quote and look for a sample size and a date. If either is missing, downgrade it from target to rough guide in your own head and in your reporting.

Check whether your two favourite sources are actually one, by comparing a few figures across them. Identical numbers to the dollar are not agreement.

Then replace the benchmark in your planning with your own ceiling, built from profit per sale and close rate.

The wider piece on judging lead cost covers where any of this fits, the 2026 figures by industry are the ones I do trust and why, and why your close rate sets the real target is the number to use instead. If you are weighing whether to hire in or out for this kind of work, the freelance and agency trade-off is an honest look from my side of it. When you want your figures checked against real sources rather than a deck, ask me.

Frequently asked questions

Are cost per lead benchmarks reliable?

Some are. The test is whether the publisher discloses a sample size and a date range. WordStream's 2026 search report discloses both, 13,474 campaigns over a stated twelve months. Several widely-quoted alternatives disclose neither, and one contains a figure that contradicts its own arithmetic.

Why do different sources publish identical numbers?

Because they are often the same numbers. HubSpot's cost per lead page reports Financial Services at $653, Legal at $649 and Ecommerce at $91, matching First Page Sage to the dollar. HubSpot says openly that it draws on First Page Sage, so quoting both is quoting one dataset twice.

What should I look for before trusting cost per lead benchmarks?

Four things: a stated sample size, a stated date range, a clear definition of what counts as a lead, and no commercial interest in the answer. A report from a company that sells the channel it says is cheapest deserves more scepticism than one from a neutral publisher.

Is there really an error in a published benchmark table?

Yes. LocaliQ lists Facebook cost per lead for attorneys and legal services at $18.17, identical to the cent to arts and entertainment. The same legal row shows a $4.10 cost per click and a 10.53% conversion rate, which works out to $38.94. Two numbers on one row disagree.

How old is benchmark data usually?

Older than it looks. First Page Sage's 2026 report states its data was collected between January 2022 and June 2025, so a three and a half year window is presented as current. LocaliQ's Facebook benchmarks were last updated in October 2025 and disclose no collection period at all.

Should I stop using benchmarks entirely?

No, just demote them. They answer whether you are in a sensible range, which is genuinely useful when you are starting out or entering a new market. They cannot answer what you should pay, because that depends on your margin and close rate, which no published table knows.

Why does this matter if benchmarks are only a rough guide?

Because people set budgets with them. I have seen campaigns paused for exceeding an industry average when they were profitable, and campaigns praised for beating one while losing money on every sale. A number treated as a target changes behaviour whether or not it deserves to.

What is the most trustworthy source right now?

For search, WordStream's 2026 study, because it publishes 13,474 campaigns and a clear April 2025 to March 2026 window. For Meta, there is no equally rigorous 2026 source, and saying so is more honest than quoting a number nobody can trace.

Want a second pair of eyes on your account?

Book a free teardown

Book a free teardown

Send me your worst-performing campaign.

I will tell you what is wrong on the first call. No obligation, no hard sell, and I will say honestly if I am not the right fit.

Book a free 20-minute account teardown

Replies within one business day, Lahore time.