Description: A content site's 6,535 article pageviews were spread across 917 URLs, yet the next-article journey remained weak. Here's how to separate successful one-page visits from a recommendation problem—and test the difference without mistaking pageviews for people.
The popular-articles panel keeps changing. Even a wonderfully obscure article about an entertainment-name mix-up makes the top ten. Surely that means readers are browsing everything, right?
Not necessarily. An article can be popular without becoming the doorway to another article. A shop can sell its star product all day while everyone heads straight back out of the door.
A busy store, a quiet corridor
In one multilingual site's Web Analytics data for October 11–17, 2026, article URLs recorded 6,535 pageviews across 917 distinct paths. The top three paths took 56.1% of those views. At the other end, 572 URLs had exactly one pageview each. A niche entertainment question unexpectedly drew 73 pageviews and made the top ten.
A separate, first-party seven-day snapshot reported on October 15 counted 2,241 article views, of which 100, or 4.5%, were internally referred views. Of 45 clicks on a “read next” recommendation, approximately 38% were followed by a return within 30 seconds.
Those two measurement systems and windows must be kept apart. The 4.5% figure is a share of internally referred pageviews, not the percentage of unique readers who opened a second article. A pageview is not a person or a session; Google Analytics likewise defines engagement using session-specific criteria rather than raw pageview counts.
Four gates between the first article and the second
First, a reason to continue. A reader who searched one question, found the answer and left may have had an excellent experience. Completion and curiosity are different states.
Second, an actually visible link. A recommendation at the bottom of a long article competes with attention focused near the top. Nielsen Norman Group's eye-tracking work found that the first screens receive a disproportionate share of viewing time.
Third, a promise worth clicking. Readers use the link label, its surrounding description and their own goals to predict what awaits. This is called information scent: a clear promise beats a vague “you may also like.”
Fourth, a destination that delivers. If someone clicks and immediately retreats, perhaps the article did not match the promise, or the page failed to load. The click was only the ticket; arriving and reading was the actual trip.
Why popularity is a poor substitute for continuity
Imagine looking up whether two comedy names refer to the same act. The site's most popular travel article might be superb, but it answers a completely different question. A follow-up such as “Why did that duo use another stage name?” preserves the thread of curiosity. That is an illustrative editorial idea, not a verified high-volume search query.
Several competing causes can produce an apparent zero: readers may already be satisfied, never see the card, dislike the suggested question or abandon the destination. The same outcome calls for different repairs.
The comparison that settles it
Compare readers exposed under the same language, topic, device and traffic-source conditions. Measure article view → card actually visible → voluntary click → successful arrival → meaningful reading. Keep automatic next-page navigation separate from hand-selected clicks; an autoplayed second page can inflate views while pleasing nobody.
Then test a popularity-ranked card against a card answering the reader's likely next question. Judge it by voluntary arrivals that turn into actual reading, with enough observations to resist a lucky afternoon.
The aim isn't to chase satisfied visitors down the street. It's to help the still-curious visitor find the next door. If someone came in for takoyaki, don't let an algorithm present a washing machine as the obvious follow-up.
The difference between zero and one more article is often the reason to keep going, not the number of articles in stock.
Sources: Nielsen Norman Group, “Information Scent” (2020); Nielsen Norman Group, “Scrolling and Attention” (2018); Google Analytics Help, engagement and bounce rate. Site figures are anonymized operational snapshots from October 2026.
