Answers
Link shortener with analytics: what the numbers can honestly tell you
Clicks, uniques, country, referrer and device, rolled up by UTC day. What each one answers, what it cannot, and why the dashboard reads rollups instead of raw events.
By Wasim Idrishi · Updated · 3 min read
Most link analytics answer questions nobody asked. Here is what the numbers are, in order of how much weight they can carry.
Clicks
A click is one request that reached the redirect and was not filtered as a bot. That is all it is. It is not a person, not a session, not a visit and certainly not a conversion. Two things inflate it and both are worth knowing about: link previews from chat apps and mail clients, which fetch the URL before a human sees it, and scanners. Preview fetchers are matched by user agent and separated out, which catches most of them and not all of them.
Unique visitors
A visitor hash, stable for one day, separates repeats from new arrivals. It is deliberately weak: no cross-day identity, no cross-link identity, nothing that could rebuild a profile. That makes uniques useful for the ratio between clicks and people, and useless for anything longitudinal, which is the right trade.
Country
Read at the edge and stored as the ISO 3166-1 alpha-2 code. Two reasons that matters. The first is that it is the same string a geo rule matches on, so the value you read in a report is the value you type in a rule. The second is that it is honest about its accuracy: this is IP geolocation, so a VPN is a different country and a mobile network is sometimes the wrong city in the right country.
Referrer
Where the browser said it came from. Increasingly empty, because apps strip it and privacy settings suppress it. When a channel matters, do not rely on the referrer. Put UTM parameters on the link and read those instead. The UTM builder assembles them without typos and the parameters survive the redirect.
Device, browser and scans
Device family separates the poster in a lobby from the newsletter. A visit that arrives through the QR endpoint is recorded as a scan rather than a plain click, so print and digital can be compared instead of being summed into one number that means nothing.
Why the dashboard is fast
Two paths. The redirect looks the link up in an in-process cache, then Redis, then the database, and returns a 30x. Click facts are pushed onto a queue and never awaited. A worker turns them into stored events and daily rollups per link.
Dashboards read the rollups only. That is why a link with a million clicks opens as fast as one with ten, and it is also why the daily numbers are UTC: a rollup keyed to a day has to be keyed to somebody's day, and UTC is the only choice that does not need a timezone attached to every row.
Retention, honestly
| Plan | Analytics detail kept |
|---|---|
| Free | 30 days |
| Pro | 365 days |
| Business and Team | 1,095 days |
Totals outlive the detail. What ages out is the day-by-day breakdown, so export anything you need for a yearly comparison before it does.
What analytics will not tell you
Whether the click became a customer. Link analytics stop at the redirect; everything after it belongs to your own site. Put UTM parameters on the link, read them in your own analytics, and use the shortener numbers for the half of the funnel it can actually see.
Related: link analytics metrics that matter, bot traffic in link analytics, and the UTC day boundary where you are.
Questions people ask
What does link analytics actually measure?
A request that reached the redirect and was not filtered as a bot. It is not a person, not a session and not a conversion. Treat it as a lower bound on interest and an upper bound on nothing.
Are clicks counted in real time?
The redirect never waits for analytics. A click is enqueued and written by a worker, so the redirect stays fast under load and the numbers appear seconds later rather than instantly.
What timezone are the daily numbers in?
UTC. Days and hours are both UTC, so if you are far from it the local evening lands on the next day in the chart. Every country page works out the exact local time your day rolls over.
Do you track people across links?
No. A visitor is identified for a day at a time by a hash used only to separate uniques from repeats. There is no cross-site profile, no advertising identity and nothing sold.