Back to field notes

GROWTH CASE 16 · TRAFFIC TOOK OFF BEFORE THE DIAGNOSIS WAS READY

A Game Shipped at 4 A.M. Reached 1M Users in 30 Days, Then Fell Off a Cliff on Day Ten

Xiaohu found a rising query, built through the night, and reported nearly 100K daily users by day four. Then traffic collapsed, leaving ads, UGC, localization, and page structure tangled in one imperfect diagnosis.

11 min read

1 public cases · 3 cited sources

Finfold Editorial

Updated ·

Public-handle account and original dashboard screenshots · the 1M users and 2.9M pageviews figures come from Xiaohu's 30-day retrospective and are not independently audited. Google guidance comes from official documentation.

Search Console CTR chart published with an international SEO game retrospective
Xiaohu's public Search Console chart shows 19,292 impressions, 25 clicks, and a 0.1% CTR on May 16. The figures are author-reported. Original Web.Cafe retrospective

At 11 p.m. on May 5, Xiaohu was looking for new queries. He checked the last 24 hours and four hours in Trends, then read the search results and saw a rising term whose leading pages did not match what users wanted to play. He bought a domain and kept building until 4 a.m. The domain also contained a typo.

He recorded 6,000 users on May 7, 16,000 the next day, and nearly 100,000 on May 9. That opening could be cut into a tidy motivational post. The second half is more useful. Supabase fell over at peak load, the main query's click-through rate reached 0.1%, and he removed ads, cleaned UGC, added languages, and changed links at the same time. Rankings later returned, but he did not pretend one change deserved all the credit.

The misspelled domain brought traffic and left a repair bill

Xiaohu noticed the domain typo only after launch. He kept the page that search had already found and used a 301 redirect toward the correct address. He says the misspelling later generated tens of thousands of visits. That outcome is too accidental to become domain-selection advice.

The reusable work happened before the purchase. He checked whether demand kept rising, whether results missed the user's intent, and whether he could ship a playable answer in hours. The query created a time window. The page still had to fulfill its promise.

When daily users approached 100,000, the database complained first

The site grew beyond one game. Xiaohu added sharing, a message wall, and leaderboards, moving from a top 50 to a top 100 and then historical and daily boards based on user time zones. Return reasons also multiplied reads and writes. Around May 9, Supabase failed. He was hiking and restarted it from his phone before addressing concurrency and abuse prevention.

The episode exposes a cost that keyword stories often omit. A new query is only the entrance. Interactive features turn a static page into a product that needs coverage, moderation, and bot protection.

The day-ten drop never produced a neat answer

Xiaohu tried Adsterra and removed it after rankings fell because the revenue was not worth the risk. A Search Console screenshot shows 19,292 impressions and 25 clicks for the main query on May 16, a 0.1% CTR. Meanwhile, usernames and phrases such as 'hours ago' repeated through the UGC and diluted the page topic.

He separated leaderboards and comments from the search document, added German, Spanish, Portuguese, and Russian pages based on user geography, and configured sitemaps and hreflang. These changes overlapped with a Google update. The term returned to roughly position 1.x around May 21, but the public record cannot isolate the cause.

A realistic expectation after the million-user headline

Xiaohu ultimately reported about one million users and 2.9 million pageviews in 30 days. An earlier GA4 screenshot shows 868,000 active users and 2.4 million views, while another Search Console image shows 1.02 million clicks. The screenshots are directionally consistent but remain author-reported.

When the main query was weakest, other languages and long-tail queries reportedly kept the site near 20,000 daily visits. A safer target for another small site is to find one intent gap, ship a useful page, watch indexing for seven days, and review query movement over four weeks. A spike may happen. It does not belong in the forecast.

Open the 1 case records and their limits

小虎

u/野生小虎 · 公开复盘中的网页小游戏

Public-handle account with dashboard screenshots

Reported result · The author reported 1M users and 2.9M pageviews in 30 days, with the main term recovering after a sharp fall

What remains uncertain · The database failed and multiple fixes overlapped, preventing a single-cause diagnosis

Open the public record

SOURCES

What each link actually supports

  1. Public-handle firsthandAccessed 2026-08-24
    Xiaohu's 30-day international SEO retrospective

    Supports the timeline, dashboard screenshots, technical failure, changes, and author-reported outcome.

  2. Official sourceAccessed 2026-08-24
    Google guidance on localized page versions

    Supports official guidance for separate locale URLs, hreflang, and sitemaps.

  3. Official sourceAccessed 2026-08-24
    Google Search spam policies

    Supports policy boundaries for user-generated spam and search manipulation.

SMALL BET, REAL SIGNAL

Run a four-week new-query test without tangling the causes

01 / SAVE INTENT

Save Trends, result-page, and user-task evidence before defining the missing answer.

02 / SEPARATE CHANGES

Change one page factor per week and log ads, UGC, and localization separately.

03 / PRICE OPERATIONS

Load-test the database, moderation, and abuse controls before a spike arrives.

CONTINUE THE WORKFLOW

BRING YOUR OWN SOURCE

Check whether your source can carry a SEO post.

Look for a person, a moment, evidence, and a consequence before generating a sentence.

Start with Finfold

More field notes

View all