A Ugly 1999 Webpage Made by an Internet Cafe Admin Sold for 50 Million Yuan: The Hao123 Story

Hello everyone, I am a rural programmer (returned to the countryside in 2020, haven't left the village since), an independent developer (committed to developing 1000 small products), and the self-styled "Frontend Tiger Chen Suiyi". My official website is: https://yicode.tech.

Today I'm not sharing a money-making story.

An internet cafe admin made an ugly webpage in 1999 and sold it for 50 million in 2004.

That webpage is called hao123.

1999, an internet cafe admin

Li Xingping, from Xingning, Guangdong, worked as an admin at a town internet cafe in 1999.

At that time, fewer than 10 million people nationwide were online, and internet access relied on dial-up telephone lines. Most people didn't even know how to write a URL.

Type one wrong letter and the page wouldn't load; they could only raise their hand and call the admin.

Li Xingping was called dozens of times a day.

Annoyed beyond tolerance, he copied down the common websites he knew, arranged them, made them into a webpage, and put it on every machine in the cafe.

It was initially called "URL Collection", later hao123 Website Home.

Technically, it was a table. No algorithms, no login, no backend. It could be finished in an afternoon.

But it worked. People who couldn't write URLs could now get online.

What he did right

Serving the people who were worst at going online

At the time, the tech circles were discussing portals and search. No one wanted to build something for "people who can't even remember URLs."

But China's new internet users were all in this group.

Completing the choice for users

Bookmarks had to be saved by yourself. A navigation site was saved for you; one click and you're off.

Not writing a single word himself

No editors. The content was all from other people's websites, so it was light, cheap, and didn't need major updates for decades.

Occupying the "first visit of the day" position

The first thing users saw when opening their browser was this page. Once a day, over three hundred times a year.

50 million, and the next business

On August 31, 2004, Baidu bought it for 50 million RMB.

At the time, the team was said to have only one or two people maintaining it for years.

Baidu was willing to pay that price because users started their search from here, and search traffic is Baidu's lifeline.

Buying the starting point means buying the upstream.

After this purchase, 265, 2345, 360 Navigation, QQ Navigation all flooded in.

Everyone knew that occupying the homepage meant collecting tolls.

Li Xingping didn't stop with the money.

In September 2004, he and Cai Wensheng created 4399 Mini Games.

Still aggregation, entry point, ads, just swapping URLs for games.

Games are more valuable than URLs because users come and don't leave; they play and then pay.

By 2008, 4399 had 300 million users.

hao123 only earned the first step's money.

What this business earns money on

There are four revenue streams.

Ad slots

Every slot on the homepage can be sold. The more front, the more expensive. Charged monthly.

Search box revenue share

Give the search box to a search engine and split based on the search volume it brings. This is the most stable.

Inclusion/recommendation

Want your website to appear on the homepage? Pay up.

Sell the whole thing

Once the entry point gets big enough, advertisers can't pay that price. The only ones who can are giants trying to hold onto traffic.

App stores, mini-program entrances, official account recommendations, AI tool rankings — in plain words, they all occupy a slot on the road users pass every day.

What you can learn

Solve your own annoyance first

Back then, he didn't think about any track; he was just annoyed enough to act.

Don't expect users to learn

It seems simple to you because you work in the industry.

Make something users open every day

A perfect tool used once a year won't grow even after ten years.

Low entry barrier doesn't mean you'll be copied

Anyone could make it, but most people stop after three months.

Once a logic works, do it again with something else

The code doesn't transfer; the experience does.

Sell when you can

50 million is getting twenty years of ad revenue in one lump sum.

Don't apply past windows to today

The 1999 window was opened by telephone lines, internet cafes, and a group of people who didn't know how to go online.

But every era spawns new opportunities; when they first appear, no one thinks much of them.

Things you can still do now

Vertical navigation

Make a list of a dozen or so websites that a specific group opens daily — designers, cross-border sellers, finance staff, village officials. The narrower, the better; narrow enough to describe their daily workflow.

First hand-write a static page, send it to twenty people, see how many are willing to set it as their homepage.

Intranet URL collection for companies

Every company has a bunch of systems scattered across documents and group announcements. New employees spend the first week asking for links.

Make a maintainable intranet navigation page and sell it to companies by the year.

The buyer is the boss; the users are employees.

New-tab extension

Web navigation requires users to open it actively. An extension occupies the new-tab page directly; open a tab and there it is.

Listing on the Chrome Web Store is cheap. Don't do homepage hijacking or those shady things.

Intelligence aggregation

Navigation sites solve "where to go"; this solves "what to see."

Bidding, policies, hiring, competitor releases — scrape them daily and push them out.

No special tech needed; what's valuable is knowing where to scrape.

Senior and children's entries

Big fonts, few items, only safe sites.

The payer is the child or the adult child; the users are old people and kids. These actually sell well.

Industry link libraries

Every industry has a set of backend addresses only insiders know.

Organize them into a page, do memberships or referral traffic.