ASN and ZIP Targeting Proxies: When You Need Them and When They Drain Your Pool

On August 6, 2026, Rayobyte added two new filters to its residential proxies — by ASN and by ZIP code. Formally, this is a minor update in someone else's office. Essentially, it’s an opportunity to delve into something that almost no one considers when purchasing traffic: at what level of accuracy are you selecting IPs and what does each additional filter cost you.

The short answer: for 90% of tasks, the country is sufficient, in another 8% you need the city, and only in the remaining percentage do ASN and ZIP truly matter. However, you always pay for unnecessary precision — through failures, IP repeats, and wasted traffic. Let’s break down how to determine your level and how to acquire ASN filtering if your provider doesn’t offer it.

Five Levels of Targeting: From Country to Autonomous System

Any residential proxy provider offers some part of this ladder:

• Country — the basic level, available to all. The accuracy of GeoIP databases here is about 99.8%.
• State / Region — for the USA, MaxMind estimates the accuracy at around 80%.
• City — about 66% for the USA, with a "hit" counted within a 50 km radius of the city.
• ZIP / Postal Code — the narrowest geographical slice.
• ASN — not geography at all, but the number of the autonomous system: a specific network of a specific operator.

ASN (Autonomous System Number) is an identifier for a network that independently announces its routes in BGP. For Comcast, it’s AS7922, for Google — AS15169, for Verizon Business — AS701. When you request “IP from AS7922,” you’re not saying “give me an address in Chicago” — you’re saying “give me the address of a residential cable subscriber of Comcast, wherever they live.”

These are fundamentally different axes. The city answers the question “where is the user from,” while ASN answers “through whom are they connected.” They are often confused, leading to convoluted configurations where someone applies three filters at once and wonders why the pool has shrunk to a few dozen addresses.

How It Looks on the Provider's Side

Technically, filters are almost always set by parameters in the proxy user’s login — a string like user-country-us-city-chicago-session-a1b2c3, to which -asn-7922 or -zip-60601 is added, depending on the specific service's syntax. The password and host remain unchanged, which is convenient: switching the level of precision is just a matter of editing one line in the config, not moving to a different infrastructure.

As of August 2026, the market looks like this: ASN targeting is available from Bright Data (country, city, ZIP, ASN), Oxylabs (country, state, city, ASN), SOAX offers a “city + ASN” combination across all plans, and now Rayobyte has joined them with ASN and ZIP at no extra cost. Rayobyte itself honestly warns in its documentation: filters can be combined freely, but each restriction reduces the number of available IPs in the pool.

This warning is the main takeaway from the entire topic.

The Arithmetic That Breaks Everything

Imagine a pool of 10 million residential IPs. The “USA” filter leaves, say, 3 million. “State of Illinois” — around a hundred thousand. “City of Chicago” — tens of thousands. Adding “AS7922” leaves only those Chicago addresses that belong specifically to Comcast. Adding ZIP 60601 — and you’re working with a pool that can measure in the hundreds of addresses, and during non-peak hours — in the tens.

What happens next if you direct a significant volume of requests at this pool:

• IP Repeats. Rotation physically cannot provide you with a new address — it starts cycling through the same ones. For the target site, this looks like one client bombarding it with hundreds of requests.
• Increased Failure Rate. Residential IPs are live devices. In a narrow slice, the share of offline nodes at any moment hits you much harder than in a broad one.
• Sessions Don’t Hold. A sticky session cannot be established if there are no free addresses in the slice.
• Traffic Costs Rise. Every retry after a ban means paid gigabytes for which you received no data.

Regarding how many addresses are actually needed for a task, we have a separate analysis with formulas — the size of the proxy pool for arbitrage and SMM. Here, it’s important to understand one simple thing: narrow targeting and a large volume of requests are mutually exclusive requirements. Choose one or the other.

When ASN is Really Needed

There are scenarios where without an autonomous system filter, you simply won’t get the needed picture:

1. Websites that respond differently to residential and server networks. Some anti-bot systems make decisions not based on geolocation but on ASN reputation. Data center autonomy gets a CAPTCHA, while cable does not. If you need to reproduce the behavior of a specific class of networks, ASN is the only lever.
2. Checking offers from telecom operators. Rates, availability of connections, upsell banners from telecom providers are shown to subscribers of their own network. From a foreign ASN, you’ll see a storefront for “new customers,” not what the subscriber sees.
3. Carrier-specific mobile scenarios. For mobile proxies, ASN is essentially the operator. Checking how an application or landing page behaves on the network of a specific mobile operator is done only this way. Here, the specialized tool is mobile proxies with their natural binding to operator networks.
4. Debugging “it doesn’t work for them.” A classic support scenario: the service is unavailable to subscribers of one provider. The problem can only be reproduced by accessing it from their ASN.

When ZIP is Needed

The ZIP code makes sense only when the business rule of the target site is tied to the postal code, not to “approximate location”:

• Local prices and availability in retail. Large retailers show prices, availability, and delivery times based on the ZIP code.
• Delivery zones and service areas. Checking “do we service this address” is a direct ZIP case.
• Verification of advertising by metro campaigns. A campaign set up for specific districts, from a country-level proxy, will show you national results.
• Insurance, utilities, real estate. Quotes and rates are often calculated directly from the ZIP code.

An important caveat: almost all of these sites still require the ZIP code to be entered manually in the interface. If there is an input field — ZIP targeting for proxies is likely unnecessary; a city is sufficient to avoid being flagged for discrepancies. The ZIP filter is needed where the site silently determines the area based on IP.

When All This is Unnecessary

A candid list of tasks where ASN and ZIP only hinder: mass scraping of directories and prices without local ties, competitor monitoring, gathering public data, country-level SEO scrapes, working with social media and multi-accounting, checking service availability by countries. Here, a broad pool with fast rotation is needed — that’s what regular residential proxies with targeting by country and city are for.

Specifically about multi-accounting: what’s needed there is not geo accuracy but stability. An account must log in repeatedly from a similar IP within the same network — this is achieved through sticky sessions, not by narrowing the pool with filters.

How to Determine Your Level: Four Steps

1. Open the target site from a clean browser without a proxy and see what it changes. If the content, price, and offers are the same for Moscow and Novosibirsk — you don’t even need the city.
2. Check if there is a manual location selection on the site. An input field for the ZIP code, a city selector, a store choice — means the site does not rely on IP as the sole source of truth, and targeting below the city level is excessive.
3. Calculate the volume of requests per hour. More than a few thousand per hour on a narrow slice — narrow targeting is mathematically excluded, no matter how much you want it.
4. Check the result based on what the site sees, not what the provider promises. This is key. The accuracy of the city based on GeoIP in the USA is about two-thirds, and “Chicago” in the provider’s panel can very well turn into “Milwaukee” in the eyes of the target resource because it looks at a different database.

We thoroughly discussed the strategy for distributing geo across large projects in the material about proxy geography for global projects — if you have dozens of countries, it’s worth starting from there.

ASN Slice on Your Own: Works with Any Provider

If your service doesn’t have an ASN filter, you can build one on your side — obtaining IPs from a broad pool and discarding those not from the desired autonomous system. You can check the address using three free methods.

Team Cymru whois — the most economical option for large lists:

• single query: whois -h whois.cymru.com " -v 8.8.8.8" will return a line like 15169 | 8.8.8.8 | 8.8.8.0/24 | US | arin | GOOGLE - Google LLC, US;
• for lists — only batch mode: one TCP connection to whois.cymru.com:43, all addresses between the begin and end lines, no more than a few thousand at a time. Team Cymru explicitly warns that those who flood the server with single queries are sent to a null route.

ipinfo.io — if you need city and ZIP along with ASN: curl -s https://ipinfo.io/8.8.8.8/json returns fields org (in the format “AS15169 Google LLC”), city, region, and postal. Convenient for quick manual checks, but free access has limits.

RIPEstat — a public API without keys: https://stat.ripe.net/data/network-info/data.json?resource=8.8.8.8 will return JSON with fields asns and prefix. It fits well into scripts when you only need the autonomous system number and prefix.

The logic of the filter is simple: take an IP from the pool → determine ASN → if it’s not the right one, release the session and request the next. Cache results by prefix, not by individual address — hundreds of IPs from the same /24 will give the same answer, and unnecessary whois requests are not needed. This approach is slower than the provider’s native filter but does not require moving and gives you full control over what you consider “your” network.

Which Proxies to Choose for Which Level

For country and city — residential proxies with a broad pool and rotation on demand; this is the workhorse for 90% of tasks. For mobile operator ASN — only mobile proxies, as data center and residential proxies simply won’t have the required autonomy. For stable binding to one network — sticky sessions; in our residential proxies, the rotation interval can be set from a change for each request to 120 minutes, which is sufficient for long multi-step scenarios.

And a sober look at the budget. According to the State of Web Scraping 2026 report (Apify and The Web Scraping Club surveyed the community in December 2025), 65.8% of specialists have started using more proxies than the year before, and 58.3% have increased their budget for them — even as the price per gigabyte in the market overall has decreased. It’s not the rates that are rising; it’s the number of retries due to increasingly complex protections. Each additional filter in the connection string adds to this bill.

Conclusion

ASN and ZIP targeting are not “more advanced proxies,” they are tools for a narrow class of tasks: telecom storefronts, local prices and delivery zones, debugging issues with specific providers, carrier tests. In all other cases, narrowing the slice only cuts the pool, increases the failure rate, and raises the traffic bill.

The practical rule: start with the country, only move up a level when you’ve proven that the target site actually changes its response. And check not the provider’s panel, but what the site sees — GeoIP lies more often than is commonly believed.