Proxy ASN targeting means choosing an exit IP by the network that announces it, on top of the country. Two US residential IPs can come from a cable provider in Denver and a hosting company in Ashburn. A defended site treats them very differently, and the country label cannot tell them apart. The network can, because every IP address belongs to an autonomous system whose owner is public. This guide covers when the network is the real requirement, how to check what you were given, and how providers differ. It rests on registry data we read on October 22, 2026. Five companies' networks announced about a fifth of the IPv4 addresses registered to the US. One of them, Charter, spread its addresses over at least nine ASNs.
What does a website look up first?
A defended site runs cheap lookups on a new connection before it reads anything you send.
Which ASN announces it
public routing data, instant
What kind of operator is that
consumer ISP, mobile carrier, hosting, cloud
What is its recent history
abuse reports, prior traffic, block lists
Only the first answer is a fact. The second is a classification the vendor maintains, and the third is reputation built up by earlier users of that address. Firewalls act on the first answer directly. The ASN match rule of AWS WAF lets a site "allow or block traffic from specific network organizations". Cloudflare's IP Access rules can "allowlist, block, and challenge traffic" by ASN on every plan.
The country appears nowhere in that sequence. The same mechanism makes the ISP proxy work: an address registered under a home provider's ASN and hosted on server hardware. The check reads the registration, so the address passes as a home connection.
How concentrated are US internet providers?
Very, at the top. We added up the IPv4 addresses that the ASNs of large US providers announced on October 22, 2026, using RIPEstat. We set them against the 1.61 billion IPv4 addresses registered to the United States in the registries' joint file of October 21. AT&T, Verizon, Comcast, Charter and Lumen together announced 331.8 million addresses, or 20.7%.
The other four fifths spread across a long tail. RIPEstat counted 32,273 ASNs registered to the US on October 21, 2026, and 18,936 of them visible in routing. For targeting, the big networks are easy to fill. A small regional provider may have few free IPs, or none, at a given moment.
Is ISP targeting the same as ASN targeting?
Not always, because one company can run many ASNs. Of the Comcast addresses we found, almost all came from AS7922. Charter, which sells under the Spectrum brand, used at least nine ASNs, most of them inherited from Time Warner Cable and Bright House Networks:
A filter on AS20115 alone covers 23.5% of Charter's addresses. Verizon spread across AS701 and two wireless ASNs, and AT&T across AS7018 and a mobile ASN. So ask a provider whether it filters by ASN number or by company name, and which numbers a name covers. A CIO Influence report of October 15, 2026, on ProxyEmpire's new ASN filter put it plainly: "one ISP often operates several".
When is the network the real requirement?
Most work does not need ASN targeting. These four jobs do. In each one, country targeting gives confidently wrong results instead of obvious errors, which makes them costly.
Ad verification
A campaign buys placements for a specific audience, and fraud often hides behind a network the campaign never meant to reach. To see an ad as a real subscriber of a named provider sees it, you exit through that provider. From another network in the same country you can see a different auction and a different ad. Our ad verification page covers the wider workflow.
Mobile carrier behavior
Carriers do things nobody else does: captive portals, filters at the network edge and zero-rated destinations. Their CGNAT maps many subscribers to one address. Reproducing any of that requires an exit inside the carrier's own network.
Testing your own detection
If you run a signup flow, a payment step or a fraud model, test how it treats a hosting ASN against a home one. That experiment across network types quickly shows whether your product quietly punishes customers of a small regional provider.
Reproducing a provider-specific bug
Peering and CDN routes differ by network. A ticket saying the site is slow only for one provider's customers cannot be reproduced from another network. An exit through that provider turns a week of guessing into one measurement.
How do you check which network a proxy uses?
Treat the pool label as a claim and the ASN lookup as the record. First learn the exit address through the proxy, then look it up:
# 1. Which address does this proxy exit from?
curl -s -x http://YOUR-LOGIN:YOUR-PASSWORD@HOST:PORT https://api.ipify.org
# 2. Which network announces that address? (our free IP lookup API, no key)
curl -s https://hproxy.com/api/ip/EXIT-IP
The second answer names the ASN, the organization and whether the address belongs to a datacenter. Here is the same lookup in our browser tool, for the public DNS address of Comcast:
The network panel is the part that matters for targeting: the ASN and its operator. A pool sold as home addresses that answers with a hosting company has told you what it is. Our IP lookup does the same in a browser, and our proxy checker adds the anonymity grade and the exit country. Our free proxy list follows the same reasoning and lists proxies by network. Two networks in one country rarely behave alike.
What do providers mean by ISP targeting?
The phrase covers at least three different products, and the documentation of each provider shows real differences.
A filter on the rotating pool
You ask for an exit from a given network, and the gateway serves one if it has an IP free there. What happens otherwise varies. PlainProxies' docs say the system "will return an error". Shifter's guide names a 502 answer when the combined filters match nothing. At Oxylabs, the docs say a country parameter overrides an ASN target, so the two cannot be combined there. Byteful's docs advise leaving the filter off unless you need a specific carrier, because it "reduces the number of available proxies".
A static ISP product
You rent a fixed address under a home provider's ASN for a period. Availability is settled up front, because the address is yours rather than drawn from a pool.
A display name over looser matching
The interface shows a provider's name, but the filter behind it is broader. One question exposes it: which ASN numbers does this name cover? A provider filtering by ASN answers with numbers.
On our side, Residential Premium takes an asn value together with a country, and our API lists the networks available per country. A target with no free IP fails by default. With strict set to false, the gateway hands out another IP from the same country instead. You can also exclude up to 20 ASNs. Our vetting guide covers how to test such claims at any provider.
How do you measure a provider's network spread?
Network diversity decides whether targeting works when you need it, and you can measure it with a small paid top-up. Pull a couple of hundred rotating exits, look up each address, and count the networks:
# 200 rotating exits; count how often each network appears
for i in $(seq 1 200); do
exit=$(curl -s -x http://YOUR-LOGIN:YOUR-PASSWORD@HOST:PORT https://api.ipify.org)
curl -s "https://hproxy.com/api/ip/$exit" | grep -o '"asn":[0-9]*,"asn_org":"[^"]*"'
sleep 0.3
done | sort | uniq -c | sort -rn
Three numbers come out of that list, and each one answers a different question.
How many distinct networks appeared
This is the real width of the pool in that country. A pool advertising millions of IPs that returns eleven networks is eleven networks wide, and every filter you apply lives inside those eleven.
What share the top network took
Concentration makes targeting fail when you need it. If one ASN carries most of your samples, a request for any other network may go unfilled.
How many exits were datacenter addresses
The lookup marks datacenter addresses. A pool sold as residential that returns cloud and hosting networks has shown its makeup in its own data.
Run the same loop against a second provider and you have a comparison built from measurements. We used the same method in our source quality study, on free proxies instead of paid pools.
Four questions worth one email
If you would rather ask than measure, these four questions are specific enough that a vague answer tells you something too.
- Which ASN numbers can I select in country X? A number is a filter. A name is a category.
- What happens when the chosen network has no free IP? An error lets you handle it. A silent substitution quietly corrupts a dataset.
- How many distinct ASNs does the pool cover there? Compare the answer with your own count afterward.
- Does targeting apply per request or per session? This decides whether a multi-step flow can stay on one network from start to finish.
What can the ASN not tell you?
The ASN answers which network announces an address. Whether a person sits behind it is a separate question, and mixing the two up leads people to over-trust the filter.
An address registered under a home provider can be hosted in a rack, which is exactly what an ISP proxy is. A home ASN can carry a compromised device, the supply problem behind the takedowns of 911 S5 and RSOCKS. And a clean ASN can carry a burned address, because reputation also lives per IP, and the ASN says nothing about the previous user.
So treat the network as the first filter of several. What sits behind it is a sourcing question. How websites detect proxies covers the checks that follow the lookup.
Which targeting level should you buy?
Finer targeting costs money and availability. Buy the coarsest level that answers your question, then move up only when a measurement forces you to.
- Country. Right for most collection work, localization checks and general access.
- State or region. Right when your subject varies inside a country: regulated pricing, regional catalogs, state-level availability.
- City. Right for local search results, delivery and store checks.
- ASN. Right when the network is the subject of the test. The four jobs above, and little else.
Our residential proxies follow that ladder. Residential Lite targets by country, from $0.50 per GB and falling to $0.23/GB at 10,000 GB+. Residential Premium adds state, city and ASN targeting, from $1.45 per GB and falling to $0.45 per GB at 5,000 GB+. You choose the targeting each time you generate a proxy line, so moving from country to ASN does not change your integration. Both run on a balance that does not expire.
ASN targeting removes one specific reason to be refused. If requests still fail at the same rate from a well-targeted home network, the problem sits in a layer this filter does not reach.
What this page could not check
- The ASNs we looked up. Company totals add up the ASNs we checked for each provider. A company can announce more from ASNs we did not query, so its true share may be higher.
- Announced against registered. Announced space is what each ASN routed on October 22, 2026. Registered space is the registries' file of October 21. The two measure different things, so the shares are approximate.
- No pools tested. We tested no provider's pool, ours included. Rival behavior comes from their own documentation, read October 22, 2026.
- Check by January 22, 2027. Routing and registrations change, and providers change their filters.
Sources
- RIPEstat routing status and AS overview for AS7018, AS20057, AS701, AS22394, AS6167, AS7922, AS7015, AS22909, nine Charter ASNs, AS209, AS22773, AS21928, AS5650, AS6128 and AS19108. RIPE NCC, read October 22, 2026.
- RIPEstat country ASNs for the United States, data of October 21, 2026.
- Extended delegated statistics of the five regional registries. Number Resource Organization, file of October 21, 2026.
- Autonomous System Number (ASN) match rule statement. AWS WAF Developer Guide, read October 22, 2026.
- IP Access rules. Cloudflare documentation, read October 22, 2026.
- ASN targeting documentation of PlainProxies, Oxylabs, Byteful and Evomi, and Shifter's ASN targeting guide of June 18, 2026, read October 22, 2026.
- ProxyEmpire adds ZIP code, ASN and OS targeting. CIO Influence, October 15, 2026.
- HProxy API documentation: residential line generation parameters asn, strict and excludeAsn, read October 22, 2026.
