INTERMEDIATE
DocsClick quality and sub-IDs
Not every click is a person. If you buy traffic or work with affiliates, some of what you’re paying for is datacenters and bots. Kilo classifies every click as it happens, so you can see the split - and, with sub-IDs, see exactly which source is sending it.
What the classification means
When someone hits one of your links, Kilo classifies that click in the same edge pass that redirects them, using network and request signals available at the moment of the click. Each click lands in one of three buckets:
Datacenter - the click came from a server network rather than a consumer connection. Real people rarely browse from a datacenter, so a high datacenter share on a source you’re paying for is a flag worth chasing.
VPN or proxy - the traffic is routed through an anonymizing network. Some of this is ordinary privacy-conscious humans, so it’s not damning on its own, but a source that’s overwhelmingly VPN traffic is worth a second look.
Likely human - the signals are consistent with a genuine person on a normal connection. This is the traffic you actually wanted.
The classification is written per click, asynchronously, independently of the redirect - so the redirect itself stays at roughly 50 milliseconds. There’s no pixel to install and no script on your destination page; the judgment happens on Kilo’s side, at the edge.
Where to read it
Open any link’s analytics in the dashboard and you’ll find the Click quality panel - the datacenter / VPN-or-proxy / likely-human split for that link over your selected time range. It sits alongside the rest of a link’s analytics: the geo, device, and referrer breakdowns you’d expect. Group analytics roll the same picture up across a whole campaign.
Read it as a shape, not a verdict on any single click. A source that’s 70% likely-human is healthy; one that’s 70% datacenter is telling you something about where that traffic came from, and probably about an invoice.
Sub-IDs: which source, exactly
A quality split for a whole link is useful, but the real question is usually which placement, partner, creative, or post sent the bad traffic. That’s what sub-IDs answer. Append a ?sub_id= parameter to any Kilo link - one distinct value per source you want to tell apart - and share those tagged variants:
qr2r.com/aA3k9?sub_id=newsletterqr2r.com/aA3k9?sub_id=partner-42
The dashboard then breaks clicks - and click quality - down per sub-ID. Instead of one number for the link, you get a row per source, and you can see at a glance that the newsletter is nearly all humans while partner-42 is mostly datacenter. That’s the difference between suspecting a source and being able to point at it.
What Kilo does, and doesn’t, block
Honest limits
Classification is always on: Kilo shows you the split on every click, on every plan - that half never changes. On top of it, Pro plans can turn on blocking for a link. It’s opt-in and off by default; when it’s on, high-confidence datacenter networks and known bots get a “verify you’re human” page instead of your destination, and everything else passes straight through the same ~50 ms redirect.
We block only what we can prove. VPN and proxy traffic stays labeled, not blocked - a lot of it is ordinary privacy-conscious people, and we’d rather show you that number than guess wrong and turn away a real visitor. So this isn’t a fraud guarantee; it’s the transparent split you can act on yourself - pausing a bad placement, questioning an invoice, rebalancing spend - plus an optional filter for the cases that are clear-cut, with both numbers still in front of you.
None of this comes at the cost of the people clicking. The classification is built from signals at click time and aggregated; Kilo never stores a raw IP address to produce it.
The last guide in this set is about putting your own name in front of all of this - running your links on a domain you own.
Open your analytics