What is a residential proxy? A residential proxy is a proxy whose exit IP address belongs to a home internet connection, so the website you reach sees a request from a household's internet connection rather than from a server. You send your request to the provider's gateway, the gateway passes it to a device on someone's home network, and that device fetches the page. Residential proxies usually rotate across a large pool and are billed per gigabyte.

This guide goes into what the short answer leaves out: where those home connections come from and why that matters, how rotation behaves in practice, what a gigabyte buys, and when a residential proxy is the wrong tool. For the side-by-side with ISP and datacenter proxies, read residential vs ISP vs datacenter proxies.

What is a residential proxy, and how does it work?

A residential request travels further than a datacenter one:

HopWhat happensWho runs it
1. Your client to the gatewayYour scraper connects to the provider with your credentialsYou
2. Gateway to an exitThe gateway picks a peer device from the pool and forwards the requestThe provider
3. Exit to the targetThe peer device opens the connection to the website from its home IPA household that opted in
4. Back the same wayThe response returns through the device and the gatewayAll three

The target only sees hop 3. Its IP lookup returns a consumer internet provider, the kind of network most of its human visitors use, and that is the point. Sites with bot protection start by scoring the network an address belongs to, and consumer networks start with a good score. Our guide to what an ASN is shows how to run that lookup yourself.

The extra hops are also the cost. The last leg runs over whatever connection the household has, so latency varies from request to request, and an exit disappears when its owner switches the device off.

Where residential IPs come from

A provider cannot buy home IP addresses from a registry: the addresses belong to consumer ISPs, which assign them to their subscribers. The only way to exit from one is to run software on a device already connected to it. So every residential pool is built from devices whose owners run something that relays traffic.

Peer apps and SDKs

Most pools today come from SDKs. An app developer, often of a free app, embeds a provider's SDK and offers users a deal: let the app use some idle bandwidth in exchange for something, such as no ads, premium features or cash. A user who accepts becomes a peer. Some providers also run their own bandwidth-sharing apps that pay users directly.

The same technology works without asking. Residential networks have been built from malware-infected machines and from free apps that bundled proxy code with no clear disclosure, and law enforcement has taken some of them down. For you as a buyer that is not an abstract ethics point:

  • Your traffic runs through those devices. If the owner never agreed, your requests are riding on someone else's compromised machine.
  • Unconsented pools are unstable. Botnets get cleaned up, apps get pulled from stores, and the exits you relied on disappear with them.
  • Your compliance review will ask. Teams that buy data increasingly have to show where it came from and how it was collected.

Good consent looks like a separate, plain prompt that says bandwidth will be shared and what the person gets in return, with a real decline button, a way to pause, and a way to leave. Payment matters because it makes the deal legible: a peer who is paid knows they are part of a network.

Questions to ask any vendor

  1. Which apps or SDKs supply the pool, and can you name them?
  2. What exactly does the peer see before they agree, and can they decline without losing the app?
  3. Are peers paid, or given something else of value?
  4. Can a peer pause or leave, and how quickly does their device drop out of the pool?
  5. Which targets does the vendor refuse, and is that list public before purchase?

A vendor that answers these in writing is easier to defend to a compliance team than one that answers with "ethically sourced" and nothing else.

How rotation works on a residential pool

Rotation is the default on residential, because the pool is made of devices that come and go anyway. A few details trip people up:

  • Rotation is often per connection, not per request. Many gateways choose the exit when a connection opens. A client that keeps connections alive, such as a requests.Session or a browser, can go out through one exit for many requests in a row.
  • Consecutive exits are unrelated. Each new exit may be in a different city, on a different carrier, with a different speed. That spreads your load, and it also means a logged-in session breaks the moment the exit changes.
  • Sticky sessions hold an exit, within limits. Most providers let you ask the gateway to keep one exit for a while, usually by adding a session identifier to the credentials. The exit still disappears if the device goes offline, so a sticky session is not a static IP. Rotating vs static proxies covers the trade-off in depth.
  • You cannot allowlist a residential exit. An address that changes, and belongs to a stranger's router, is no use to a partner's firewall.

Per-GB billing: what a gigabyte buys

You do not rent an address on residential; you share a pool. The provider's cost is the traffic it moves through other people's connections, so residential is billed per gigabyte of traffic, whichever exits carried it.

That makes the size of your requests the number to know. curl reports the bytes it downloaded (-w write-out variables):

curl -s -o /dev/null --compressed \
  -w "body %{size_download} bytes, headers %{size_header} bytes\n" \
  https://en.wikipedia.org/wiki/Proxy_server

When we ran it on 2026-09-29 it printed:

body 59901 bytes, headers 6415 bytes

About 66 KB for one compressed HTML page, so roughly 15,000 pages like it fit in a gigabyte. Three things shrink that figure. TLS handshakes and your own request headers add overhead, and providers differ in whether they count both directions. A headless browser that loads the same page also fetches its images, scripts and fonts, which can multiply the bytes many times over. And every retry after a block costs the same bytes as a success.

So measure your own targets before you buy a volume tier. Residential proxy pricing explained turns these numbers into a monthly cost and a cost per successful request.

When a residential proxy is the right tool

  • Protected targets at volume. Retailers, travel sites, search engines and marketplaces that score hosting networks harshly, where you need many unrelated exits. See proxies for web scraping.
  • Wide, shallow collection. Many independent requests for public pages, each small, none of which needs to remember the one before.
  • Sampling what ordinary visitors see. Ads, prices and page variants served to consumer networks rather than to servers.
  • When datacenter IPs have already failed. If a fair test on datacenter addresses gets blocked at a rate you cannot live with, residential is the next step.

When a residential proxy is overkill

  • The target accepts datacenter traffic. Many APIs, public data sources and smaller sites do. A datacenter proxy is cheaper per request and faster.
  • You move a lot of bytes per request. Media, file downloads and full browser renders make per-GB billing expensive quickly.
  • The target needs to remember you. Logins, carts and account work need one address that stays put. A static ISP IP keeps a consumer network's reputation without the rotation.
  • A partner allowlists your IP. Rotation breaks allowlists by design.
  • You need a specific location at ProxyHive. Our residential line has no location selection today; ISP and datacenter are placed by country, region and city.

Residential proxies at ProxyHive

ProxyHive residential is a rotating pool billed per GB, starting at $5.50/GB, with volume tiers on the residential pricing page. What you should know before buying:

  • No location selection today. Traffic rotates across the whole pool with no country, state, city or ASN choice. If a country matters, use ISP or datacenter.
  • Traffic never expires. A gigabyte bought today is still there next year, and rates are locked for 12 months after purchase.
  • Sub-users. Each can have its own credentials and its own share of your traffic, so one job cannot eat another's allowance.
  • Connection details live on the order. The dashboard at app.proxyhive.io lists the host, port and credentials for your order; copy them from there. For session control, see the options on your order in the dashboard.

New accounts get 1 GB of residential traffic to test against their own targets first. Every use has to fit the allowed-use policy, which is published before you pay.