Most articles about short link analytics are invented. They hand you a clean pie chart, tell you "78% of users are on mobile," you redesign your landing page around it, and nothing changes.
This one is different. Below is the raw data from five months on PicKu: 27,013 clicks across 120 countries, from May 25 to October 21, 2026. Every number below comes from real records, including the unflattering parts.
The totals: scale decides what you can do
806 short links, 27,013 total clicks, 27,002 detail rows. That 11-row gap is not a bug. Most short link systems skip the detail row when a crawler or security scanner hits the link (it redirects, but carries no referer and no complete user agent), or when the JS redirect hasn't rendered yet when the bot fetched it. That is 0.04% and can be ignored. If your gap exceeds 1%, though, look at your redirect implementation.
The second thing worth noting: 120 countries over five months averages 225 clicks per country, and the vast majority of countries contributed between 1 and 10. So when your dashboard proudly reports "120 countries," keep in mind that 90 of them together account for a rounding error.
Referrer is where you fool yourself most easily
| Referrer | Clicks | Share |
|---|---|---|
| (direct, no referrer) | 13745 | 50.9% |
| a partner site (owned) | 8198 | 30.4% |
| this site (owned) | 2420 | 9.0% |
| another owned site | 1441 | 5.3% |
| security.feishu.cn | 275 | 1.0% |
| link.zhihu.com | 155 | 0.6% |
| www.google.com | 139 | 0.5% |
| www.facebook.com | 46 | 0.2% |
First, why "direct" is over half. This isn't a broken statistic; it's the fate of short links.
The main use of a short link is stuffing a long URL into a chat window. When someone taps it from WeChat, QQ, or email, the browser gets either no referrer or the referrer of the app's in-app container. Modern apps largely use their own WebViews, and many strip the Referer header outright as a privacy measure. That is not something you can intercept.
So: a high "direct" share is normal in every short link system's referrer breakdown. If a tool claims 100% source tracking, it is either lying or normalizing only the tracked subset into something that looks clean but isn't usable.
Now the uncomfortable half. The three self-owned domains above total 12,220 clicks, or 45.2%. Nearly half of your "analytics data" is traffic you sent yourself.
That traffic isn't fake, and it isn't invalid. But it means if you read this dataset as "where do my users come from," your own promotion channels will completely mislead you. You would conclude that your users come from your other site, then spend double on that site, closing a self-congratulatory loop.
The fix is to separate self-owned sources from external ones. Strip ours out and the real external natural traffic here is 714 clicks, 2.6% of the total. Don't dismiss it: 0.5% Google conversion traffic is genuinely good for a stable utility site.
Devices: the 23% "Unknown OS" is the real story
| OS | Clicks | Share |
|---|---|---|
| Windows 10 | 9360 | 34.7% |
| Unknown OS | 6320 | 23.4% |
| Android | 5109 | 18.9% |
| Mac OS X | 2800 | 10.4% |
| iPhone | 2424 | 9.0% |
| Linux | 645 | 2.4% |
| iPad | 125 | 0.5% |
| Windows 7 | 91 | 0.3% |
My first reaction to Unknown OS in second place was "the parser is broken." It isn't. Three things land in that bucket: app WebViews that omit key fields or disguise themselves outright; in-app browsers like WeChat's and QQ's (27.8%, covered below); and search engine prefetching, since Baidu, 360, and Bing crawlers hit short links constantly.
The third one deserves emphasis: short link pages are among the most heavily crawled page types on the web.Any public short link shared into a group chat will be crawled within hours. Counting those as analytics is measuring bot traffic.
At least half of that 23% is WebView plus crawlers, both of which distort the device dimension. That is why "mobile share" is such a soft number across most short link tools.
One solid conclusion does survive: Windows 10 alone at 34.7%, plus 91 on Windows 7, makes desktop Windows the absolute majority. So if your short links point at landing pages with images, popups, or splash screens, design for 1366×768 through 1920×1080 first, mobile second. That runs counter to most people's instinct, but the data decides it.
Browsers: "Handheld Browser" at 27.8% means WeChat
| Browser | Clicks | Share |
|---|---|---|
| Chrome | 12632 | 46.8% |
| Handheld Browser | 7520 | 27.8% |
| Unknown Browser | 5359 | 19.8% |
| Firefox | 790 | 2.9% |
| Safari | 695 | 2.6% |
| Internet Explorer | 5 | 0.02% |
| Opera | 1 | — |
"Handheld Browser" is not a real browser. It is the fallback label a server-side UA parser assigns when it cannot identify the engine at all. Everything in that bucket is, in substance, a mobile WebView. I cross-checked this: the bucket's 27.8% share points the same direction as mobile devices being over half of all traffic, and its frequency sits far above any genuinely obscure desktop browser (Opera managed just 1 click).
For anyone building a landing page, this matters far more than "Chrome is 46.8%":
- WeChat's in-app browser doesn't support all modern CSS — backdrop-filter, :has(), and some position: sticky behavior.
- WeChat intercepts pages with outbound links, so an outbound button may render as an interstitial.
- WebP support depends on the WeChat version, and older WebViews can show a blank block where you carefully optimized an image.
So if you run short links in China, always open your splash page inside WeChat itself rather than trusting Chrome's device emulator. I've seen far too many cases that "looked fine in the emulator but showed no image in WeChat."
Countries: a blank is not "unknown country," it's "not resolved"
| Country | Clicks |
|---|---|
| china | 10127 |
| united states | 6174 |
| france | 1722 |
| (blank) | 1417 |
| japan | 1268 |
| united kingdom | 875 |
| germany | 609 |
| hong kong, China | 441 |
| canada | 420 |
| taiwan, China | 399 |
| south africa | 388 |
| republic of moldova | 385 |
That blank 1,417, or 5.2%, doesn't mean "from an unknown region." It means the request carried no resolvable IP. Common causes: server-side internal redirects, proxy requests, some CDN origin fetches, and crawlers.
Here's the easy misreading: lots of people see "China 10,127, US 6,174" and conclude half their users are in China and half in the US.But with 13,745 direct clicks and a blank referrer column, there's no way to tell how many of those were Chinese users clicking Chinese pages versus overseas users clicking Chinese links.
One obscure but telling signal: Republic of Moldova at 385 clicks ranks 12th. Moldova has roughly 2.5 million people. A country of 2.5 million contributing 1.4% of all clicks, behind Japan, the UK, and Germany, is not a normal number.
My read is that this traffic isn't Moldovan users at all, but overseas VPS or proxy egress IPs that geolocate to Moldova, or simply imprecise IP database data for that region. Moldova is a known proxy-relay hub in Europe.
The lesson isn't "localize for Moldova." It's always distrust the outliers in the long tail.When a country suddenly enters your top five, check IP attribution and traffic authenticity before you build a campaign for it.
Dates: a 207-per-day average reveals a weekly rhythm
October 12–21 totals 2,068 clicks, a daily average of 206.8. Weekdays averaged 266; the weekend days (Oct 12 and 13) managed only 97 and 169. The weekend collapses to 40–60% of weekday volume, and the pattern is very stable.
That's directly useful for scheduling:
- Weekday daytime is the main arena — that's when load time and reliability matter most.
- Low weekend volume isn't the same as low weekend value.Short link shares tend to be sent on weekends and clicked on weekdays. My sample doesn't support that conclusion yet, though; more days of data would be needed to confirm it.
- Never judge a trend from a single day.Oct 13's 169 looks like a drop, but it's a Saturday. That's a misjudgment I nearly made myself.
So what is all this data actually for
If you finished those five tables and thought "so what," here are three things I actually do with them.
1. Decide landing page dimensions instead of guessing. Desktop Windows at 34.7% means design for 1440px wide and fold down for mobile, not the reverse. Most people get this order backwards.
2. Decide what your stack must support. Handheld Browser at 27.8% means old WebView compatibility is mandatory: no :has(), don't lean on backdrop-filter, ship both WebP and JPG.
3. Strip self-owned traffic before deciding where to spend. Remove your own domains and external natural traffic is 2.6%. That number is small, but it tells you something more important: you currently have almost no external acquisition. Growth money belongs in external channels, not in doubling down on your own sibling site.
What these numbers do not tell you
An analysis that only reports good news is worthless, so here are the limits:
- Short links can't track what happens after the click. You learn the origin and the destination, nothing in between, unless you own the destination too.
- Referrer data is inherently incomplete. As above, that 50.9% blank share is structural, not a bug to be fixed.
- IP geolocation has error. VPNs, proxies, and datacenter IPs map foreign users to local ones and vice versa.
- Crawlers are mixed in. Short link pages are crawler magnets, and how many of those 27,002 rows are humans cannot be cleanly separated.
- Only clicks are counted, not impressions. A short link nobody clicked simply does not exist in this dataset.
The takeaway
The valuable part of short link analytics isn't "which country is biggest." It's knowing where you can't see.
In this dataset, half of 27,013 clicks have no visible source, 23.4% have no identifiable OS, 27.8% have no identifiable browser engine, and 5.2% have no resolvable country. Quantifying what you can't see is worth far more than pretending it's all "unknown users" and optimizing blindly.
One footnote: those 27,013 clicks came from PicKu's own analytics panel. All four dimensions — referrer, device, browser, country — are raw counts straight from the database, with no normalization or smoothing applied. If you want to see the same picture for your own links, create a free account, run a few links through it, and come back to compare numbers. https://picku.org
If you run a short link tool, run the same query (SELECT os, COUNT(*) FROM stats GROUP BY os and the like). You'll probably get a chart that differs quite a bit from your intuition. Come compare numbers.