Aug 10, 2026
Which Link Metrics Actually Mean Something
Link analytics give you more numbers than decisions. Most people look at total clicks, feel something, and change nothing. Here's which figures actually support a decision and which are just weather.
Total clicks: a volume number, not a performance one
Total clicks tells you how much traffic happened. It tells you nothing about whether it was good traffic, or whether anything you did caused it.
It's only meaningful compared against itself over time, and only when the other variables hold still. Comparing this week to last week is useful. Comparing a link you shared once to one you shared eight times is meaningless.
Unique vs returning: the ratio is the signal
The split between unique and returning visitors is more informative than either number alone.
- Mostly unique — you're reaching new people. Good for growth, expected for a campaign.
- High returning — the same people keep coming back. Good for reference content, a documentation link, a menu. Suspicious for a one-off offer, where it often means a small group clicking repeatedly rather than an audience.
A sudden shift in this ratio without a change in what you're doing is worth investigating. It usually means the traffic source changed.
Bot share: the number that fixes every other number
If your tool separates bot hits, look at that figure first, because it recalibrates everything else.
Every share into a messaging app or social platform triggers preview crawlers. A link posted in several group chats can accumulate a dozen automated requests before a human sees it. If bot hits are a large fraction of the total, your "clicks" are mostly infrastructure, and any conclusion drawn from the raw number is wrong.
This is also the honest explanation for why your numbers never match a platform's.
Time of day and day of week: the most actionable, least used
This is where a decision actually lives. If your clicks cluster at 7am and 9pm, that's when your audience has their phone out — so that's when to send. If a link posted at 2pm gets nothing and the same link at 8pm gets traffic, you've learned something you can apply this week.
Very few people look at this, and it's the report most likely to change what you do tomorrow.
Geography: useful mainly as a mismatch detector
Country data matters less for congratulating yourself and more for spotting a mismatch. If most clicks come from a country you don't ship to, price for, or write for, that's a real problem hiding inside a healthy-looking total.
City-level data is coarse and derived from network information — treat it as a region, not an address.
Referrer: mostly missing, and that's expected
Most social and messaging traffic arrives with no referrer, because in-app browsers strip it. A large "direct" bucket is normal and not a failure of your tracking.
Which is precisely why one link per placement does the work referrer data can't. If you want to know which channel delivered, the answer has to be built into the link, not extracted afterwards. See how links behave on social platforms.
Device: check it, then act on it
Device split is only interesting if you do something about it. If 80% of your clicks are mobile and your destination is a desktop-first page with a PDF on it, that's the finding — not the percentage.
The comparison that beats all of them
Clicks per placement, over time, for the same destination.
One link in the newsletter header, one in the footer, one in the Instagram bio, one on the flyer — same destination, four links. That comparison is the only one that isolates the variable you control. Everything else is influenced by things you don't.
A short weekly routine
- Check bot share, so you know what the other numbers are worth.
- Compare this week's clicks to last week's, per placement.
- Look at time of day once a month and adjust when you post.
- Find the placement with the fewest clicks and stop doing it.
That last step is the one people skip, and it's where the time is.
See these breakdowns on your own links — sources, devices, countries and timing, free, with bot filtering on by default.