How a growing site can help visitors find their next read without making them work for it
Short answer: the problem is usually not the number of articles. It is that the site leaves all the work of looking to the reader. Show a few picks, then always offer the next direction, and let the site suggest what to search for.
More articles should mean more choice. For the reader, it often feels like the opposite.
You open the homepage. Same stories as yesterday. You come back a few hours later. Same stories again. You go to the full list and spin the mouse wheel. Titles fly past. You can tell there is a lot of stuff. You can't remember a single thing you saw.
So someone says, "Fine, let's show only six." Now anyone who doesn't like those six leaves on the spot, before they ever see number seven.
Add a search box as the last line of defense, and it still doesn't help much: people who are just browsing usually don't have a search term in mind.
The articles are there. Search is there. Recommendations are there. And still nobody finds anything.
This is not a content shortage. It is a site that makes the reader do all the hunting.
In this article, the homepage, the article list and search are treated as one system for helping people find things, not as three separate parts.
0. The 30-second version: don't shrink the store, make the shop assistant more helpful
A huge list is tiring. So you show only six. That is half right.
But if the visit ends when those six miss, you have removed the list's weak point and also the lucky finds it used to give you.
What you need is:
a small set of picks, plus the site suggesting where to go next.
- "Not this? How about a real-life story?"
- "Want to compare options?"
- "Want to know how it works?"
- "Something lighter?"
- "A totally different topic?"
Give people the next step. Don't just ask them, "What do you want to search for?"
The site should take half a step forward and say, "Is this what you're looking for?"
1. Why the homepage starts to feel like "this again": new to the site is not the same as new to you
Research on online news recommendations has studied "user fatigue": people lose interest when they are shown the same thing again and again.[1]
In a study of real Bing Now logs, 15 news items were shown at a time. When the same person came back within two hours, an average of 11 of them were repeats. When the ranking also took into account how many times an item had been shown and how the person had reacted, results improved by up to 15% for all users and by up to 34% for frequent users.[1]
The point is not "delete old articles."
A three-month-old article is new to someone who has never seen it. A ten-minute-old article feels like "this again" if that person has already seen it four times today.
So what matters is:
not just how new the article is, but how new it is to this reader.
Keep track of how many times each card was shown, when it was last shown, whether it was clicked, and whether it was actually read.
Make unseen items easier to show. Lower an item a little after it has been ignored several times. After someone reads an article, show other articles near that topic.
But don't turn something like "push it down after two views" into a universal rule. Measure on your own site after how many views people stop reacting, and set the number from that.
2. When the list stops being information and turns into a waterfall
Make the list page long enough and, in theory, everything can be seen.
In practice, a mouse wheel or a finger flick sends it rushing by.
At that moment the list is no longer a set of cards you can compare. It is a waterfall of text.
If there are 100 articles but only 8 really get noticed, the other 92 are just sitting in storage.
It is also too simple to say "more articles means people can't choose."
A 2010 review that pooled 50 experiments with 5,036 people found that, on average, having too many options had almost no effect, but the results varied a lot from study to study.[2]
In other words:
having a lot is not the problem. Making people compare a pile of options that all look the same, all at once, is.
A list's job is not to show everything. It is to make sure people always know where they are and where they can go next.
3. "Just show six" is not the finished answer: what about people who didn't like the six?
A small set has real advantages. It fits on one screen and is easy to compare. It survives fast scrolling. Each card can look different from the others.
But the weakness is clear:
if all six miss, people leave before they ever meet number seven.
So think of six not as the answer, but as one window to look through.
Next to it, put:
- another six
- more real-life stories
- more practical
- lighter
- go deeper into this topic
- something completely different
In a two-week live experiment on the news site NU.nl in 2026, a method that kept picks relevant while making them less similar to each other raised the click rate and made people feel the picks were more relevant.[3]
This doesn't mean you can just alternate categories.
Variety doesn't mean random. It means "not too alike, but still interesting."
4. Don't delete the full list: move it from the front door to the back room
You still need the full list.
It is great for people who want to find something they read before, dig into one category, browse everything, or check articles by date.
But handing the full list to someone who just came by thinking "anything fun here?" is too much.
Instead of "choose from 127 articles," say "out of 127, here are a few that look good right now."
Keep the full list as a path for heavy users: "see all," "narrow it down," "sort by date."
The place where people discover things and the place where you keep everything don't have to be the same page.
5. The biggest weakness of a search box: it makes the reader come up with the words
Site search is handy. But it only works if one thing is true:
the person can put what they want into words.
Most people have learned that Google works even if they type something rough.
A small site's search box hasn't earned that trust yet.
On top of that, someone killing time often doesn't know what they want yet.
So an empty box that just says "Enter a keyword" doesn't get anything started.
In Baymard's large-scale tests of how people use search, users often used the suggestions under the search box not just to send as they were, but as a starting point for writing their own search.[4]
So the value of suggestions is not only saving keystrokes.
It is helping with "What should I even search for?"
6. Thoughtful search: suggest what people might mean, not just article titles
If someone types "work," show:
- I hate the work itself
- the people are wearing me out
- I'm not sure whether to quit
- I want to work faster
- I want to read someone's real experience
These are not article titles.
They are guesses at what the person is trying to find.
Even before anything is typed, people can start from:
- What should I do?
- Which one should I pick?
- Why does this happen?
- How did it actually go?
- Got anything fun?
And right after someone finishes an article, offer:
- "If you want to compare next"
- "If you want to see how it works"
- "If you want the opposite example"
The moment the search box opens, suggestions based on what you just read are already there.
Reader: "Ki ga kiku nee!" — roughly, "Wow, that's thoughtful."
That line is a Japanese internet meme that spread widely in 2026 through a tapioca video by the cosmetic surgeon and YouTuber Mikiya Takasu.[5] Another streamer has also claimed to be the original source of the phrase, so here we only call it "a meme that spread through Mikiya Takasu."
The meme isn't the point. The point is building something that gets that reaction:
before you type anything, the next move is already a little bit ready.
That is what thoughtful search means.
7. You don't need to call AI every time someone types to make search feel thoughtful
There is no need to run an expensive AI writing tool on every keystroke.
When you create an article, create the information for finding it at the same time.
For each article, save:
- related search words
- the kind of problem it helps with
- nearby topics
- topics that look similar but are different
- its tone: light, practical, or deep
- questions a reader is likely to have next
When someone visits, just pick a few items using:
- the current article
- articles shown recently
- articles clicked recently
- a boost for things not seen yet
- variety within the list
AI helps write these suggestions while the article is being edited. The live site uses a cheap way of sorting them.
That is fast, costs little, and makes it easy to see why a suggestion appeared.
8. Don't build the homepage, the list and search as separate islands
A homepage that rotates fast. A list split into pages. A really smart search.
Even with all three, if they don't talk to each other, readers start from zero every time.
What they should share is:
what this person saw recently + what worries or topics each article answers + how many times each one has been shown.
On the homepage, offer "How about this today?" On the list, show an amount people can actually compare, plus where to go next. In search, show suggestions even when the box is empty, and help with what people mean while they type.
Connect all three with one shared picture of where the reader is.
9. If you only watch the click rate, you end up with a clickbait machine
Clicks went up. Success.
If that is all you measure, the most over-the-top headlines win in the end.
Look at the whole chain:
the article is picked as a suggestion → it actually shows up on screen → it gets clicked → the page opens → it gets properly read → the next article → the person comes back
For this problem, also look at:
- how people react to the same card the 1st, 2nd and 3rd time
- how many items overlap with the last visit when someone returns
- how often "show me others" is used
- how often people search from a suggestion
- how often they edit a suggestion a bit before searching
- how often a search finds nothing
- how often people leave after fast scrolling
- how often people keep looking even when they didn't click anything in the first set
- how often articles that were never shown before get some exposure
The most important one:
when the first six missed, did the reader make it to number seven?
10. Conclusion: from "please search" to "is this what you're looking for?"
Every growing site eventually runs into a storage problem.
Show everything and it rushes by. Show a little and it misses. Add search and nobody types. Fix the recommendations and people get bored. Personalize too much and everything gets narrow.
Looked at one by one, these seem to clash.
But give each part its own job and it sorts itself out.
The stock can be big. The screen can be small. The next directions can be many. The site can write half of the search for the reader.
Putting six cards in place of a huge list is not the answer.
Put a thoughtful next step behind those six.
That is when a big number of articles stops feeling like "too much" and starts feeling like "so much to discover."
Checklist before you build it
- Can you record how many times the same card was shown?
- Do you treat "new article" and "new to this reader" as different things?
- Are there several ways forward when the small set misses?
- Did you keep the full list and just give it a different job?
- Can people start exploring from an empty search box?
- Do suggestions help with what people mean, not just with spelling words?
- Can related search words and next questions be saved when an article is created?
- Do you avoid treating "not clicked" as "not interested" right away?
- Do you look at reading, the next article and return visits, not just clicks?
- Do articles that have never been shown get a turn?
- Are numbers updated from your own site's data instead of being treated as the one right answer?
References (5)
- Hao Ma, Xueqing Liu, Zhihong Shen (2016), User Fatigue in Online News Recommendation doi.org
- Benjamin Scheibehenne, Rainer Greifeneder, Peter M. Todd (2010), Can There Ever Be Too Many Options? doi.org
- Robin Verachtert, Kim Falk, Christine Bauer (2026), Enhancing Diversity in News Recommendations Increases Click-Through Rates doi.org
- Baymard Institute (2024), Always Copy the Active Autocomplete Suggestion to the Search Field baymard.com
- Public posts and video listings about the meme, checked in September 2026. Who first said it is intentionally left open
