Bot Traffic IndexThe bot problem, quantified

Residential Proxies: How Bot Traffic Learned to Look Local

A store's analytics show a spike in traffic from a residential neighborhood, on real devices, with real browsers, at plausible hours. The traffic is bots. The IPs belong to real homes, the ASNs belong to real internet providers, and every network check the store runs comes back clean, because the network checks were built for a world where bots came from datacenters, and that world is gone.

The datacenter giveaway is gone

Bot detection used to be simple at the network level. Bots came from datacenters, and datacenters have ASNs and IP ranges that are easy to list. A request from a datacenter IP, in a city like Ashburn or Boardman or Council Bluffs, was a bot until proven otherwise. The check worked because the economics favored it: datacenter bandwidth was cheap, residential bandwidth was not, and bot operators used the cheap option. Residential proxies changed the economics. They route bot traffic through IP addresses assigned to real homes by real internet providers, and the traffic looks like it comes from a residential connection, because it does. The server logs versus analytics gap is the same problem from the measurement side: the traffic the analytics sees and the traffic the server sees are different sets, and the gap is where the bots live.

What residential proxy traffic looks like

It looks like a customer. Real ISP, real city, real device fingerprint, plausible hours, plausible session length. The signals that used to work, ASN, IP reputation, geolocation, all pass, because the network origin is genuine. The bot is not spoofing a residential IP. It is using one, rented from a network that pays real homeowners for their bandwidth or that routes through compromised devices. The traffic is indistinguishable from a customer at the network level, and the store that relies on network checks will not see it at all. The GIVT versus SIVT split is the right frame: general invalid traffic is the kind a list can catch, and residential proxy traffic is sophisticated invalid traffic, the kind that needs analysis, not a list.

Why the old checks fail

ASN alone fails, because the ASN belongs to a residential provider. IP reputation fails, because the IP belongs to a real home, and blocking it blocks a real customer. Geolocation fails, because the location is real. The datacenter city spike, Ashburn, Boardman, Council Bluffs, The Dalles, Dublin, Singapore in the top cities, was the most reliable tell in GA4, and residential proxies erase it, because the traffic now comes from the same cities the customers do. The checks did not get worse. The traffic got better at looking like the thing the checks were designed to pass.

What still works

Behavior. Residential proxies hide the network origin, but they do not hide what the traffic does: the speed of the session, the pattern of the requests, the consistency of the fingerprint, the anomalies in the session. A bot that visits forty product pages in ninety seconds is a bot, whatever its IP. A session with no mouse movement, no scroll, and a purchase in twelve seconds is a bot. Detection moves from where the traffic comes from to what it does, and the signals that work are behavioral, multi-point, and corroborated, the same standard the MRC sets for sophisticated invalid traffic. The network check is the first filter. It is no longer the last one.

Frequently asked questions

What is a residential proxy?

A residential proxy routes traffic through IP addresses assigned to real homes by real internet providers, instead of through datacenter servers. The traffic looks like it comes from a residential connection, because it does. Bot operators rent access to these networks and run their bots through them.

Can I still block bots by ASN?

No, and that is the point. An ASN check identifies datacenter traffic. Residential proxy traffic comes from residential ASNs, the same ones real customers use, so the check passes. ASN alone was never a complete defense, and residential proxies are why.

What still works against residential proxy bots?

Behavior. Residential proxies hide the network origin, but they do not hide the behavior: the speed, the pattern, the session anomalies, the fingerprint inconsistencies. Detection moves from where the traffic comes from to what it does, and the signals that work are behavioral, not network-based.

Can a real customer's device be part of a botnet?

Yes, and it is a separate problem from detection. A real customer's device that has been compromised becomes part of a botnet, and its traffic looks exactly like the customer's own traffic, because it is coming from the customer's device. The only defense is behavior: the compromised device acts differently from its owner.