What Is a Proxy? How It Works, Types, and Uses
You start using a tool, script, or bot, and the instructions tell you to connect it to a proxy. They do not really explain why. Choosing one raises more questions: some proxies are sold individually, others in packages, prices vary, and the descriptions are full of unfamiliar terms. Before buying anything, you want to know what you need proxies for and how to choose the right ones.
What is a proxy?
A proxy is an intermediary that your software, script, or bot uses to reach websites. Your software connects to the proxy, the proxy contacts the website, and then it carries the response back.
Behind a proxy is a computer or another device connected to the internet, running software that passes data between the two sides. This device is often called a proxy server. The proxy reaches websites using an IP address that you gain access to when you buy the service.
Why your software might need a different IP address
Your internet connection has an address — an IP address. A website sees it when your bot connects directly. When your bot uses a proxy, the website receives the connection from the proxy’s IP address.
A website may use an IP address to estimate a visitor’s country or limit the number of requests from one address. The IP your software uses can therefore affect its work with that website.
For example, several bots running on one computer normally reach the internet through the same IP. If each needs a separate address, proxies can provide that arrangement. Another task may need an address associated with a particular country or the ability to change IPs during work. Those requirements help determine which proxies to choose.
What you receive when you buy proxies
Buying proxies gives you access to one or more IP addresses for your software. When your bot connects through the service, websites see those addresses instead of your usual internet connection’s IP. The company that supplies this access and keeps the service running is called a proxy provider.
The provider gives you a proxy address and a port. For example, 192.0.2.40:8080 contains the address 192.0.2.40 and the port 8080. A port is a number that identifies which service to connect to at that address — in this case, the proxy.
Access may also require a username and password. The provider supplies these with the address and port, and you enter them in your software’s settings. The illustration below shows which value belongs in which field.
Once configured, the bot can reach websites through the proxy. Other apps on the computer continue using their own connection settings. For example, the bot may use a proxy while your browser uses your usual internet connection. Buying proxy access does not switch anything by itself: you need to configure it in the software that will use it.
How does a connection through a proxy work?
To see what changes after setup, compare a website connection with and without a proxy. This applies to both a browser and a bot: each sends requests to a website and receives responses. A browser shows a page to a person; a bot processes the returned data according to its instructions. Either can illustrate the connection route. We will use a browser below.
Opening a page without a proxy
You enter a website address in your browser. The browser connects to the website and asks it to send a page. That is a request. The website sends a response, and the browser displays the page it receives.
The website sees your internet connection’s IP address. If a bot on the same computer contacts the website without a proxy, it will normally use that same IP.
Adding a proxy
Now you enter a proxy in the browser’s settings and open the website again. The browser connects to the proxy and tells it which website it needs. The proxy contacts the website, receives the response, and passes it back to the browser. The website sees the proxy’s IP instead of your connection’s IP.
The same happens with a bot. It continues doing its work, such as publishing a prepared post or checking for new messages. The proxy carries its requests and the website’s responses. The bot still determines which actions to perform and what to do with the returned data.
Why one connection address can provide access to several IPs
A provider offers many IPs but gives you just one address to enter in your bot. That does not necessarily limit you to one IP: the address in the settings can lead to one entry point for the service.
For example, the provider gives you proxy.example:8080. This connection address is written as a name, although it can also be supplied as a numeric IP address. You enter it in your bot’s settings along with the port. This entry point into the proxy service is called a gateway.
The bot connects to the gateway, and the service contacts websites using other IPs that you have access to. The diagram shows two possible outgoing addresses behind one gateway: IP A and IP B.
We will explain later how an IP stays the same or changes.
Why use proxies?
People use proxies to hide their real IP address from a website, use an IP from a particular country, or give a bot one IP to use throughout its work.
These options are possible because the proxy connects to the website on your behalf. The website sees the proxy’s IP instead of yours and uses that address to identify the connection’s country. Which IP each bot gets, and how long it uses that IP, depends on the proxies and settings you choose. With the right setup, you can do all three at once.
Collect data from websites
Suppose your bot tracks prices in an online store and checks thousands of products. A regular visitor would not open that many pages in a short time. If every request comes from one IP, the website may detect automated collection and block that address. The bot can no longer retrieve the data.
Proxies let you spread the bot’s requests across different IPs: it checks some products through one address, some through another, and so on. Each IP sends fewer requests. This helps get around the website’s limit on requests from a single address and keep collecting data.
Access websites as a visitor from another country
The same online store may show different products, prices, and delivery options to visitors from different countries. Some websites also block access from IPs in your country.
With a proxy in the country you need, the website sees a local IP instead of yours. This lets you check what the store offers local shoppers or get around a country-based access restriction.
Keep a stable address for continuing work
Some websites allow connections only from public IP addresses added to an allowlist. A long-running process may also need to continue using the same address across many requests. A proxy with a stable assigned IP can provide that address even when the customer’s own internet connection changes, provided the customer can still access the proxy.
Keeping the same IP is useful here. Replacing the address after every request would work against the purpose. The program’s own login state and data still need to be maintained; keeping one IP does not itself preserve a website account session.
Give separate processes separate addresses
If you run several bots or separate tasks, you can assign a different proxy IP to each. This separates their website requests and makes it easier to see which address each task uses.
Different tasks depend on different proxy features. Regional observation depends on location; continuing work depends on address stability; separate workloads depend on how addresses are assigned and shared. The rest of the guide explains these properties so that words such as “residential,” “private,” and “rotating” can be read as distinct facts about a service.
Where do proxy IP addresses come from?
When choosing proxies, you will come across four types: datacenter, ISP, residential, and mobile. They differ in where their IP addresses come from and which connection your bot uses to reach a website.
Datacenter proxies run on servers in data centers and use IP addresses from hosting companies. The website sees a connection from a server network.
ISP proxies also run on servers in data centers, but use IP addresses from home internet providers. The equipment is server hardware, while the address belongs to a network that serves ordinary internet users.
Residential proxies send requests through users’ home internet connections. Your bot reaches the website through a device on a home network and uses that connection’s IP address.
Mobile proxies send requests through mobile internet, for example through a smartphone or a modem with a SIM card. The website sees an IP address from a mobile operator.
This matters when choosing a proxy because websites may treat datacenter, residential, and mobile IPs differently. If a website restricts connections from data centers, addresses from home internet providers or mobile operators may help you work with it.
For a closer comparison of the four types, their costs, and how they work, see Proxy Network Types.
Who can use the same proxy IP?
When buying proxies, you need to know whether you will be the only person using an IP or whether other customers will receive it at the same time. This is the difference between shared and dedicated proxies.
Shared proxies let several customers of the provider use one IP. Your bots and other customers’ software reach websites from the same address. If someone sends too many requests and a website blocks the shared IP, your bot will also be unable to access that website through that address.
Private, or dedicated, proxies give you exclusive use of an IP for the period you pay for. The provider’s other customers do not use it, so their activity does not consume the website’s request allowance for your address.
Shared proxies usually cost less. Dedicated proxies are useful when you want to use an IP on your own without depending on what other customers do through that address. For more help choosing, see Shared and Dedicated Proxies.
How does an IP address stay the same or change?
Some tasks need the same IP for weeks or months. Others need frequent address changes. This is why proxies are divided into static and rotating types.
A static proxy lets you use one IP for weeks, months, or longer. It suits tasks where a bot or account needs a stable address over a long period.
A rotating proxy changes the IP while your software runs. For example, each new request can use a different address. Another option is to change the IP at set intervals, such as every 10, 20, or 30 minutes. The bot keeps working, while the website receives its requests from different IPs.
To change addresses, the service uses an IP address pool: a set of addresses available to it. When it is time to change, the service selects an IP from that set.
How to keep one IP temporarily with a rotating proxy
Sometimes a bot needs to complete several actions in a row using one IP: for example, sign into an account, open the required page, and submit a form.
A sticky session temporarily keeps one IP for these actions. Depending on the service, you may be able to keep an address for 10, 20, or 30 minutes, for example. The provider sets the available durations and the conditions for retaining the IP.
This lets a rotating proxy change addresses between tasks while keeping one IP throughout each task. For more on configuration, see Static Proxies, Rotation, and Sticky Sessions.
How is a proxy’s location determined?
Websites commonly estimate location from databases that associate IP ranges with countries, regions, cities, and networks. This is IP geolocation. It is an interpretation of an address, rather than a direct measurement of the visitor’s physical position. Different databases can return different results for the same IP, and their records can change over time.
There are several places involved in a proxy connection: where your device is, where the gateway is hosted, where traffic reaches the destination from, and where the destination’s database places that IP. Those places need not all be the same. A nearby gateway can provide access to addresses associated with another country. Its proximity may affect the first part of the connection without determining the location the website sees.
A provider may let you choose a country, a city, or a particular network. City-level labels are also estimates, so the name in a control panel is not a guarantee of how every website will classify the IP.
The countries, cities, and networks you can choose depend on the service. A pool containing addresses from several countries does not necessarily let a customer choose one country. Likewise, a country selector does not imply that every city or operator within that country is available. For regional work, the relevant property is the location the destination associates with the actual address being used.
Language, saved preferences, account settings, and device-location permissions can still influence what the page shows. IP location supplies one part of that decision. Proxy geolocation: country, city, and network explains how those signals and location options relate to one another.
How does a program communicate with a proxy?
A protocol is a set of rules that both sides use to communicate. The program must know how to ask the proxy to reach a website, and the proxy must understand that request. An address and port identify where to connect; the protocol determines what the program says after reaching it.
With an HTTP proxy, the program uses HTTP proxy rules. For an ordinary unencrypted HTTP request, the proxy can receive a request identifying the destination and forward it. For an HTTPS website, a common method is the HTTP CONNECT request: the program asks the proxy to open a tunnel to the destination, then carries the browser-to-website encrypted exchange through that tunnel.
The label HTTPS proxy needs context. It can mean that the connection from the program to the proxy is itself protected with TLS, the encryption protocol used by HTTPS. Some offers use the label to describe support for HTTPS destinations through CONNECT. These are separate capabilities. Encryption between the browser and website and encryption between the program and proxy protect different parts of the route.
SOCKS5 provides a more general way for a program to ask an intermediary to communicate with a destination. It can identify a destination by IP address or domain name. The standard includes TCP connections and a mechanism for UDP traffic, but a particular provider or program may support only part of those capabilities. SOCKS5 itself does not encrypt the transferred content; an application can use its own encryption through the connection.
The destination’s protocol and the proxy protocol need not have the same name. A browser can reach an HTTPS website through an HTTP proxy tunnel or a suitable SOCKS5 service. What matters is whether the program and proxy support the required arrangement.
The protocol does not determine the network type or whether the IP changes. For example, an ISP proxy can support SOCKS5 and keep a fixed address, while a residential proxy can support HTTP and change addresses. The reverse combinations are also possible: ISP proxies with HTTP and rotation, and residential proxies with SOCKS5 and a fixed IP. Network type, protocol, and IP rotation can be combined in different ways.
HTTP, HTTPS, and SOCKS proxy protocols explains the communication rules and what they mean for different applications.
How does a proxy control access?
A paid service needs a way to distinguish customers who may use it from other internet users. Authentication lets the proxy service check whether a connection is allowed, using the supplied credentials or source IP address.
With a username and password, the program supplies credentials when connecting. The provider can use them to identify the customer, apply a plan’s limits, and sometimes select a location or session. The proxy password belongs to the proxy service. It is separate from a password used to sign into a website through that connection.
With IP allowlisting, the provider accepts connections from public source addresses that have been approved in advance. It checks the address your connection presents to the proxy. This is usually different from a private address used inside a home or office network. If the public source address changes, an authorization based on the old address may stop matching. If several devices share an approved public address, an address-only rule does not distinguish those devices individually.
Some services authenticate you with a username and password. Others allow connections from IP addresses added to an allowlist. Some require both: enter the username and password in your software, and add your public IP to the allowlist in the provider’s control panel.
This check protects access to the proxy. It does not make a shared IP dedicated, hide your account login from the website, or encrypt data by itself. For more on setting up both methods, see Proxy authentication methods.
How do IPv4 and IPv6 proxies differ?
An IPv4 proxy contacts a website from an IPv4 address, such as 192.0.2.40. The website sees that address. It is written as four numbers separated by dots.
An IPv6 proxy contacts a website from an IPv6 address, such as 2001:db8::40. Here, the website sees a different address format: it uses digits and letters from a to f, with groups of characters separated by colons.
Compatibility with your target website matters when choosing. Some websites accept only IPv4 connections; others accept both IPv4 and IPv6. To reach a website from an IPv6 address, the website must support IPv6.
IPv6 proxies therefore cannot automatically replace IPv4 proxies for every website. For more on choosing and configuring them, see IPv4 and IPv6 proxies.
Why Elite proxies do not guarantee access to a website
A service description or a proxy checker may show Elite / High Anonymity. This means the proxy does not add request headers that disclose your original IP address or identify the use of a proxy. The label describes one property of the proxy, not its overall quality.
The website still sees the proxy’s IP. If the address appears in a database of known proxies or has a poor reputation, the website may reject requests from it. Browser settings and software behavior can also affect the website’s decision. Elite therefore does not mean a proxy cannot be detected, and it does not guarantee access.
Proxy Anonymity Levels explains Elite and the other labels in this older classification, along with what anonymity checkers actually test.
Does your plan provide enough capacity for your task?
Customers want to get more work done for less money. Providers pay for servers, IP addresses, network connections, and maintenance; they need to cover their costs and make a profit. Each plan therefore defines how much capacity the provider offers for its price.
When choosing proxies, compare these terms with your software’s workload. Even a large IP package can be unsuitable: there are enough addresses, but the program gets errors, runs too slowly, or uses up its paid traffic before finishing the job.
| Limit | What to check before buying | Possible consequences |
|---|---|---|
| Number of allowed concurrent connections | How many connections your software opens at once, and how many the plan allows. | New connections are rejected when the allowance is already in use. Some tasks fail to start or end with an error; retries increase the time needed to finish the work. |
| Number of allowed requests per second or minute | How often your software sends requests, and what rate the provider permits. | Requests above the limit are rejected or delayed. The program receives fewer responses within the available time, and retries slow the work down further. |
| Maximum data transfer speed | Whether the available speed is enough to transfer the required amount of data within the available time. | Transfers reach the speed limit, so files take longer to download. If the package has a shared speed limit, adding parallel downloads does not increase it: the downloads share the available speed. |
| Paid traffic allowance | How many gigabytes the package includes, and how many your software will need. | Once the allowance runs out, work stops until you top up or the package renews. If paid overages or automatic top-ups are enabled, work may continue, but costs increase. |
Check whether each limit applies to one IP, the whole package, or the account. If a limit applies to the whole account, buying more IPs does not increase it by itself. And “unlimited traffic” does not mean unlimited speed or any number of concurrent connections.
Besides limits, response time matters: how long your software waits for the start of a response. Long waits make even small actions slow. If a response does not arrive in time, the program may end the request with a timeout error, meaning that the allowed waiting time has expired. A ping to the proxy alone does not show how long it will take to get a response from your target website.
Providers may restrict destination ports, websites, applications, or types of activity. A port restriction can prevent a particular service from being reached even when the proxy is available and customer authentication succeeds. These restrictions belong to the proxy service. A destination can separately reject or limit incoming requests according to its own rules.
Proxy usage limits explains how these limits affect your software’s work.
What does the customer pay for?
Proxy costs can depend on the number of IP addresses, the amount of data transferred, or the number of requests. The same software workload can therefore be billed differently under different plans.
When paying per IP, you buy access to a certain number of IP addresses for a set period, such as 100 IPs for a month. This model is common for proxy lists assigned to a customer. During the paid period, each new request does not increase the charge for the addresses themselves. The plan specifies whether traffic is included and whether its volume is limited.
When paying per GB, you pay for the amount of data transferred through the proxy. A short text response uses little traffic, while images, videos, and large files use much more. The more data your software transfers, the faster it uses up the paid allowance.
When paying per request, costs depend on the number of requests the service counts for billing. Check what it treats as one request and whether errors and retries are charged. For example, opening one page can involve separate requests for images and other parts of the page.
The payment arrangement is a separate choice. A subscription provides a package for a set period with regular renewal, such as 100 IPs or 50 GB each month. Pay-as-you-go means paying as you use the service: for example, money is deducted from your balance for traffic consumed. Check the chosen plan for expiry and unused-allowance rollover rules.
To compare offers, estimate your workload: how many addresses you need, how much data will be transferred, and how many requests your software will make. For more, see Proxy Pricing Models.
What to check before buying proxies
Before paying, find out what happens if the proxies do not suit your task, you need to replace the addresses, or your paid access expires.
Can you test the proxies first? Check whether there is a free or paid trial, how long it lasts, and which proxies you will receive. A trial lets you check that they work with your software and the website you need before buying the main package.
Can you get a refund? Check when the provider refunds payments and how much time you have to ask. In particular, find out whether a refund is available if the proxies work but do not suit your website or application.
Can you replace the IP addresses? Find out when replacements are allowed, how much they cost, and how often you can request them. For example, can you replace an address that works but is blocked by the website you need?
What happens at renewal? Will you keep the same IPs or receive different ones? If you need to keep your addresses, find this out in advance. Also check whether the service renews automatically and what happens if you do not pay on time.
What help does support offer? Find out where to turn if the proxies will not connect or the website you need will not open through them. Check what the provider offers if the problem cannot be resolved.
Read the page for your chosen plan and the service terms for the specific conditions. For more on what to look for, see Proxy provider policies.
How to tell whether proxies suit your task
The right proxies depend on what you need to do. One task may require keeping the same IP address for weeks; another may need frequent changes. One program may need just a few connections, while another opens thousands. A service that suits one task may therefore be unsuitable for another.
Understanding proxy characteristics helps you see what a provider offers, which limits matter to you, and what to clarify in advance. Testing with your software and the websites you need will show how well the service works for your actual task.