Every serious click-fraud discussion starts with datacenter filtering, and for good reason: it's cheap, it's local, and it catches a meaningful share of low-effort bots. If a click's ASN organization resolves to a known hosting provider, you can flag it before it ever touches your database, at zero marginal cost.
But datacenter filtering only catches traffic that never bothered to hide. The traffic that actually costs a network money runs through residential proxy networks — real household IP addresses, rented out through consumer apps and SDKs, indistinguishable from a genuine visitor at the ASN level. A residential proxy IP belongs to a real ISP, in a real neighborhood, because it is one.
That's the problem no local blocklist solves. It requires a live, crowdsourced reputation signal — a database of "this specific address has been seen behaving like a proxy exit point recently" — which is fundamentally a different kind of data than "this ASN belongs to Amazon." It's the one place in our fraud pipeline where we reach for an external signal, and we reach for it narrowly: only after the free local checks have already run, cached aggressively, and never as the sole reason to block a click outright.
The honest framing: no single layer solves fraud. A network that tells you otherwise is selling a slogan, not a system.
Want a network that works like this?
Free to join, reviewed manually, transparent from the first click.