One Article Earns Everything

01689 articles, 20 that matter

I run blogs on a lot of small sites, and I had a feeling that each one was carried by a single article. Feelings are cheap, so I pulled 90 days of Google data, to 25 October, for every site I can read in Search Console. Twenty seven of them had at least three articles that Google showed to anyone. Between them, 689 articles were shown at least once and earned 5,526 clicks. Then I sorted them. The single best article took 24.4% of every click. The top five took 53.4%. The top twenty took 78.9%. A blog is not a library that earns evenly. It is one or two articles carrying everything else.

Most articles earn nothing

Four hundred and seventy eight of the 689 articles, 69.4%, never earned a click in three months, even though Google showed each of them at least once. Only 51 earned ten clicks or more. This is not one bad site dragging the numbers down. Take away the largest blog and the other 26 sites still put 52.9% of their clicks on their single best article each. A voice product had 38 articles shown and one of them earned 389 of 761 clicks. A CV product had 29 and one earned 92 of 108.

02What the earners have in common

One plain question, answered completely

The top articles on these sites are not the most ambitious ones. One answers how much of a settlement a person actually keeps. One lists the words that make a CV read as written by a machine. One compares the free tools in its category, the product's own tool among them. Each answers a question people type in plain words, and each answers it completely on one page, with the number or the list near the top. The article that earns is the one whose title is a question somebody actually types. Studies, essays and announcements, the pieces I was proudest of, mostly sit in the silent 69%.

Not quality, demand

It is tempting to read this as good articles and bad articles. It is mostly about demand. Several of the silent articles are careful and correct; nobody searches for what they answer. That is the same finding as the thin pages that earn, where a templated set I planned to delete outperformed the pieces I liked, and as calculators get quoted, where pages that compute something for the reader were clicked five times as often as explanations.

03Feed the winner, then find the next question

Treat the winner as a product page

If one article carries half of a blog, it deserves product treatment: keep its facts current, give it a dated last checked line, link to it from the homepage and the articles around it, and make sure the next step a reader might want is one click away. Do not rewrite it for the sake of it. A page that ranks is fragile, and the safest improvements are the ones that add without removing. The best use of a new article is often to send people to the old one that already works.

Choose the next topic from searches, not ideas

Before writing anything new, look for the next plain question with real demand, in your own Search Console queries and in tools that show search volume. The silent 69% are mostly topics chosen because they were interesting to write. The same split runs through assistants, which quote the explainer or list that answers a common question far more than the essay, the pattern in write explainers, get cited by the listicles. Interesting to write and useful to find are different lists, and only one of them earns.

The limit of this data is worth stating. Search Console only lists pages it showed at least once, so articles that were never shown at all are missing from the 689, which makes the real silent share higher, not lower. And clicks are not citations: an article can be quoted by assistants without a single click. For this blog the numbers are blunt. Twenty nine entries were shown in those 90 days and none earned a click, which is why its record and the agency's guides now play different roles.

Tomorrow: a search for next year that already has an audience, and one that does not exist yet.