The Kissmetrics Blog: 62M Views, $15M, and What the Data Says

A.The scale

The blog ran from 2008, hit its stride around 2011, and was effectively finished by 2018. Across the posts we can measure, it took about 62 million pageviews. Google sent 49.2 million of those hits from 34.3 million people; every other search engine combined sent under 715,000. It collected 96,728 social shares, two thirds of them on Twitter.

Signups peaked in 2015 at 47,265 and revenue peaked in 2014 at $4.6M. Those two peaks being a year apart is the first useful detail: the blog was filling the top of the funnel for a year before the money showed up, and it kept sending signups for a year after revenue began falling.

B.The concentration nobody plans for

589 posts acquired at least one paying customer. Forty-five posts carried half of all attributed revenue, and 173 carried eighty per cent. The remaining 1,600-odd posts split what was left, and 598 of them, a third of everything published, produced 15.9 million views and not one paying reader.

That is a more extreme distribution than the usual 80/20 hand-wave. It means the blog’s entire commercial output could have been produced by about two months of writing, if anyone had known in advance which two months.

The honest reading is that nobody did know, and the volume strategy was how they found out. But it also means the *maintenance* of 1,600 low-yield posts was a cost carried for years against no return, and that cost never appeared in any content report, because content reports measure publishing and traffic rather than yield.

A.The category that brought the most traffic earned the least

Sorting every post by subject and dividing revenue by views produces the finding that makes the rest of this article worth reading.

SEO content took 12.1% of all views and returned 6.2% of revenue. Analytics content earned nearly seven times more per view on a quarter of the traffic. Social media took 3.5% of views for 2.6% of revenue.

The search logs say the same thing from the demand side. Of 268,635 non-brand search hits, 24.1% were social media queries. The single biggest non-brand term in the blog’s history was “how to get more likes on facebook”, at 12,263 hits. Analytics queries were 5.2% of demand and produced a sixth of all revenue.

A blog for marketers at large will always find its largest audience on social and SEO, because that is where marketers at large are searching. The problem is that an analytics company does not sell to marketers at large. It sells to the much smaller group with a measurement problem, and that group was arriving through a different door entirely.

B.Engagement was not a quality signal, and following it made things worse

The natural instinct is to let the audience tell you what to write more of. The data says that instinct, applied to shares, points in the wrong direction.

Across 1,085 posts with 3,000 or more views, shares per 1,000 views against first-touch revenue per 1,000 views correlate at r = +0.028, and comments at r = −0.044. Measured on raw counts instead, shares and revenue correlate at +0.288, which is not a finding: views drive both, and views against shares is +0.503. The rate version is the one that answers whether a post shared more than its traffic predicts also earns more than its traffic predicts. Both are noise. The top 106 posts by shares produced $690,980 of first-touch revenue. The top 106 by revenue produced $6,857,738. The two lists overlap by fifteen posts.

The title words that predicted sharing were lessons (2.8× baseline),likes (2.7×), hacking (2.6×) and growth (2.4×), which is, almost exactly, the vocabulary of the 598-post tier that produced no customers at all. Optimising for engagement did not merely fail to help. It actively steered the editorial calendar toward the content that never converted.

Structure fared no better as a predictor. Word count, paragraph length, sentence length, heading density, bullets, images and outbound links were all correlated against both engagement and revenue. Every coefficient came back under 0.16. Top-decile and bottom-decile posts by sharing are structurally near-identical: 17.5 versus 18.0 median sentence words, four bullets each, 3.8 versus 4.0 images per thousand words.

C.What did predict revenue was framing

One thing separated cleanly, and it was not style. It was what the title asked the reader to do.

The honest reading of that table is narrower than it first looks, and it is worth being explicit about which parts survive scrutiny. Comparison at 9.05× rests on six posts and one of them is an outlier large enough to set the bucket by itself. Contrarian at 2.40× rests on fifteen. Both are hints.

The one result with enough posts behind it to argue from is the numbered list at 0.68×, across 298 posts and 21 million views. Listicles were the single most published format on the blog and the worst-earning one, by a margin that holds however you slice the sample.

And the format everybody now warns against comes out fine. “How to” scored 0.97×, which is the baseline.

The clearest evidence that format is the wrong lever sits at the very top of the revenue table. The five highest-earning posts in the archive are all about Google Analytics, and between them they use four different title formats: two descriptive, two how-to, and one numbered list. The listicle among them earned $278,814. If format were the driver, that post should not exist.

What they have in common is not shape. It is that each one meets a reader inside a specific, painful problem with the analytics tool they already have. A task-shaped title is not the problem; a title with nothing at stake is.

This makes sense once you look at who is searching. Someone typing “how to calculate LTV” has already decided what to do and wants the arithmetic. Someone typing “is LTV worth tracking” or landing on “most LTV formulas are wrong for subscription businesses” is still deciding, and a decision is the moment a product can enter the conversation.

A.What Neil Patel bought

In 2019 the Kissmetrics domain went to Neil Patel. Todayblog.kissmetrics.com issues a 301 to neilpatel.com/blog/, and article-level URLs redirect to his blog homepage rather than to equivalent articles.

On paper this is one of the better content acquisitions of the era. He acquired a domain with a decade of accumulated authority, tens of thousands of referring links, and a corpus that had ranked for a very large number of commercial marketing queries. Redirects pass most of that equity. For a business whose product is marketing advice and whose audience is marketers at large, the fit is close to exact: the highest-traffic categories in the archive were Marketing, SEO and Social Media, which is precisely his territory.

He bought, in other words, the 15.9 million views that never converted for an analytics company. For him they are not the dead tier. They are the audience.

B.What Kissmetrics lost, and it is not the traffic

The obvious loss is the traffic and the backlinks, and they are genuinely gone: the redirects live on a domain the company no longer controls, so republishing the content elsewhere cannot recapture the link equity. Anything rebuilt starts from zero.

The more interesting loss is subtler. The blog was worth $15 million because it was joined to the product. Every post could be traced to the people who read it and the money they paid, because the blog and the app shared an identity layer. The content was valuable; the measurement is what made it manageable.

When the domain left, the corpus and the measurement were separated. What Neil Patel holds is a very large amount of traffic without the join. What Kissmetrics kept is the join without the traffic: six years of behavioural and billing data showing exactly which subjects, framings and formats produced customers.

Of the two halves, the second is the one you cannot buy. Traffic is purchasable and rebuildable. A six-year record of which content produced revenue, at the level of the individual post, is not available to anyone else in this market, including the people who now own the pages.

C.What we are doing with it

We ran the full analysis against our current blog and found that of the 173 posts carrying 80% of historical revenue, 149 have no equivalent today. That is $9.9 million of proven first-touch revenue in subjects we are simply not covering, and 97% of it is still evergreen: checkout abandonment, why A/B testing usually wastes time, why pageviews and time-on-site are the wrong goals, cohorts, pricing pages, mobile conversion loss.

We are rebuilding those, in that order, ranked by what they earned. Not republishing the 2014 text, since the arguments need current mechanism and current data, but answering the same decisions for the same reason.

And we are not repeating the measurement mistake. Traffic reporting is the reason a quarter of that blog could produce ten million views and nothing else for years without anyone raising it.