Proxies are powerful — and easy to misuse. This guide covers the concepts, vocabulary, and decisions you need to understand before spending a cent on proxy services.
- The fundamentals
How a proxy actually works
A proxy is a middleman server that sits between your computer and the website you’re visiting. Instead of the website seeing your IP address, it sees the proxy’s IP address.
| You |
Middleman |
Target |
| Your computer |
The proxy server |
The website |
| Sends a request. Your real IP is never exposed to the target. |
Receives your request, forwards it using its own IP address. |
Sees the proxy’s IP, not yours. Responds back through the proxy. |
**Why does this matter?**
Every device connected to the internet has an IP address — a number that identifies where the request is coming from. Websites use this to enforce rate limits, geo-restrictions, and bot detection. A proxy gives you a different IP, so you can appear to be coming from a different location, bypass per-IP limits, or simply stay anonymous.
The core concept is simple, but there are several layers of vocabulary you’ll need to understand before choosing a proxy. They all connect back to this basic idea.
- IP behaviour
Static vs. rotating IPs — the first decision
Before thinking about what kind of proxy to buy, you need to decide whether you need a fixed IP or one that changes with every request.
Static
Same IP every time
You always connect through the same IP address. The website recognises you as the same returning visitor.
Good for: managing accounts, consistent logins, tasks where you need to appear as a single user over time.
Rotating
Different IP per request
Each request (or each session) goes through a different IP address. The website can’t track you across requests.
Good for: scraping, price monitoring, bypassing rate limits — any task where volume matters more than identity.
Sticky session
Same IP for one session
You keep the same IP for the duration of a task (e.g. a checkout flow), then rotate to a new IP for the next task.
Good for: multi-step flows like adding to cart, logging in, and checking out — where a mid-flow IP change would trigger detection.
Common beginner mistake
Using rotating proxies for account management. If your account logs in from a different IP every time, most platforms will flag it as suspicious activity and lock the account. Use static or sticky session proxies for anything account-based.
- Proxy types
Datacenter, residential, ISP, and mobile — what’s the difference?
The type of proxy determines where the IP address comes from, and therefore how much a website will trust it. This is the most important decision you’ll make.
| Type |
Where the IP comes from |
Detection risk |
Speed |
Cost |
| Datacenter |
Rented servers in data centres (AWS, OVH, etc.) |
High — sites block datacenter IPs aggressively |
Fastest |
Cheapest |
| ISP |
Real ISP IP ranges, but hosted in data centres |
Medium — looks residential, performs like datacenter |
Fast |
Mid-range |
| Residential |
Real home internet connections (peer network) |
Low — looks like a normal user |
Slower |
Expensive |
| Mobile |
Real 4G/5G mobile connections |
Lowest — carriers are trusted by default |
Variable |
Most expensive |
The key insight
Websites don’t just check whether your IP is “good” — they check what kind of IP it is. A datacenter IP is easy to identify because all IPs from that range belong to the same cloud provider. A residential IP looks like a real person’s home internet, so it’s far harder to block without also blocking real users.
Which type do you actually need?
| Your task |
Recommended type |
Why |
| Scraping public data, APIs with low protection |
Datacenter |
Speed and cost matter more than stealth |
| Sneakers, limited releases, retail drops |
ISP or Residential |
These sites hard-block datacenter ranges |
| Social media accounts |
Residential |
Platforms fingerprint IP type aggressively |
| Geo-restricted content or streaming |
Residential |
Most services block known datacenter IPs |
| Highest-protection targets, app platforms |
Mobile |
Carrier IPs are treated as trusted by default |
- Performance concepts
Bandwidth, speed, and latency — the numbers that matter
Proxy providers advertise a lot of numbers. Here’s what they actually mean and which ones you should care about for your use case.
B. Bandwidth
The total amount of data transferred through the proxy, measured in gigabytes (GB). Most residential and mobile proxy plans charge per GB. If you’re scraping small amounts of text, you’ll use very little bandwidth. If you’re loading full web pages with images, it adds up fast. Always estimate your bandwidth needs before buying a plan.
L. Latency
How long a single request takes, measured in milliseconds (ms). Datacenter proxies are typically under 100ms. Residential proxies can be 500ms–2s because the request routes through a real person’s home internet connection. For most scraping tasks, latency doesn’t matter much. For time-sensitive tasks like sneaker drops, even 200ms can be the difference between success and failure.
T. Throughput (concurrent connections)
How many requests you can make at the same time. If you need to scrape 10,000 pages quickly, you’ll want to run many parallel requests. Providers have limits on concurrent connections per plan — check this before buying if you’re planning high-volume work.
R. Rotation interval
For rotating proxies, how frequently the IP changes. Options typically range from “per request” (new IP every single time) to “every 10/30/60 minutes.” Per-request rotation is best for scraping. Longer intervals suit tasks where you need some session continuity but still want IP diversity over time.
✓ Practical tip: estimate your bandwidth before buying
Average web page: ~2MB. Average API response: ~5KB. If you’re scraping 10,000 web pages, that’s ~20GB of bandwidth. At $10/GB for residential, that’s $200 just in bandwidth. Switch to an API or mobile-optimised version of the page, and that same task might cost $2–5.
- Risk and compliance
Legal and terms-of-service risks
Using proxies is legal in most countries. What you do with them determines whether you’re on safe ground. Here’s a clear breakdown — but always consult a lawyer for your specific situation, since laws vary by country and context.
| Generally safe |
Grey area — check first |
| Accessing publicly available data |
Scraping sites that explicitly prohibit it in their ToS |
| Price comparison and market research |
Creating multiple accounts on a platform |
| Bypassing geo-restrictions on content you own access to |
Buying limited-release products at scale (sneakers, tickets) |
| SEO monitoring and rank checking |
Automated social media activity |
| Ad verification and brand protection |
Bypassing paywalls or login walls |
Illegal in most jurisdictions
- Accessing systems you’re not authorised to access (CFAA in the US, Computer Misuse Act in the UK)
- Scraping personal data without consent (GDPR in Europe)
- Using proxies to commit fraud or circumvent fraud detection
- Credential stuffing and unauthorised account access
**
Terms of service violations ≠ illegal, but consequences still exist**
Breaking a website’s terms of service is not a criminal offence in most countries, but it can result in account bans, IP blocks, or in some cases civil action. The bigger risks for most proxy users are practical (getting blocked) rather than legal. That said, large-scale scraping of certain data types — especially personal data or copyrighted material — can cross into illegal territory quickly.
- Getting started
Your first steps
Before you buy anything, answer these four questions. They’ll determine which proxy type, pricing model, and provider is right for you.
| Q1 |
What are you actually trying to do? |
|
Be specific. “Scraping” means different things at different scales. “Sneakers” on Nike is different from a smaller boutique. “Social media” depends entirely on the platform. The more specific your answer, the more precisely you can match to the right proxy type — and avoid overpaying for capabilities you don’t need. |
| Q2 |
How sensitive is the target site to proxies? |
|
Test it first with a free or trial proxy before spending money. Sites like Google are fairly tolerant. Sites like Nike, Ticketmaster, or Instagram have sophisticated bot detection and will block cheap proxies instantly. Knowing this upfront saves you from buying the wrong tier. |
| Q3 |
What volume are you running? |
|
Estimate: requests per day × average response size = bandwidth per day. Then multiply by 30 for a monthly estimate. This tells you whether per-GB or per-IP pricing makes more sense, and how large a pool you’ll need to avoid overloading individual IPs. |
| Q4 |
Do you need to appear as one user or many? |
|
If you’re managing accounts or doing anything session-based, you need consistent IPs per account (static or sticky). If you’re doing volume tasks where individual IP identity doesn’t matter, rotating proxies are more efficient and cost-effective. |
| ✓ |
Always test before you commit |
|
Every reputable proxy provider offers a trial or small test allocation. Run 100 requests on your actual target before buying a month’s subscription. If the provider doesn’t offer a trial, that’s a red flag — it usually means their IPs won’t perform on your target and they know it. |
Test what you’ve learned
Three quick questions. Select an answer to see whether you’re right — and why.
| You’re managing 20 social media accounts. Which proxy type should you use? |
| Datacenter — it’s faster and cheaper. Static residential, one IP per account. Rotating residential — more IPs means more safety |
| A website says “200 countries, 50 million residential IPs.” What should you check before buying? |
| Whether the price is competitive. How many countries are covered, the IP count in your specific target countries, and whether they actually verify as residential |
| You’re scraping a major e-commerce site for product prices — 50,000 pages per day. What matters most? |
| Lowest possible latency, pool size, success rate, and cost per GB, sticky session reliability |
- Reference
Glossary
Every term you’ll encounter when researching or using proxies, in plain language.
| IP address Core |
| A numerical label (e.g. 192.168.1.1) assigned to every device on the internet. Websites use your IP to identify where you’re connecting from and to apply rate limits or blocks. |
| Proxy server Core |
| A server that forwards your internet requests. The website sees the proxy’s IP address instead of yours. |
| Rotating proxy |
| A proxy setup that automatically assigns a different IP address to each request or each session. Useful for high-volume tasks where you need to avoid per-IP rate limits. |
| Static proxy |
| A proxy that always uses the same IP address. Essential for account management where consistency of identity matters. |
| Sticky session |
| A rotating proxy configured to keep the same IP for the duration of a single session or task, then rotate to a new IP for the next session. |
| Residential proxy |
| A proxy that routes your traffic through real home internet connections. The IP appears to belong to a regular household, making it much harder for websites to detect and block. |
| Datacenter proxy |
| A proxy running on servers in data centres (cloud hosting). Fast and cheap, but easily identified and blocked by sites that check whether an IP belongs to a cloud provider. |
| ISP proxy |
| Proxies that use IP ranges assigned to internet service providers, but hosted in data centres. They appear residential to sites but perform more like datacenter proxies in terms of speed. |
| Mobile proxy |
| Proxies routed through real 4G/5G mobile connections. The highest trust level — mobile carrier IPs are rarely blocked because blocking them would affect real mobile users. |
| Bandwidth |
| The total amount of data transferred through a proxy, measured in gigabytes. Most residential proxy plans charge per GB of bandwidth used. |
| Latency |
| The time it takes for a request to travel from your computer, through the proxy, to the website, and back. Measured in milliseconds. Lower is faster. |
| P95 latency |
| The latency at the 95th percentile — meaning 95% of requests are faster than this number. More useful than average latency because it shows how bad the slowest requests get. |
| IP pool |
| The total set of IP addresses available to rotate through. A larger pool means each IP is used less frequently, reducing the chance of any single IP getting blocked. |
| IP reputation |
| A score assigned to an IP based on its history. IPs used for spam, bot activity, or scraping accumulate a bad reputation and get blocked by more sites. Providers maintain “clean” IPs by rotating them regularly. |
| Geo-targeting |
| Choosing proxies in a specific country or city so that you appear to be browsing from that location. Used to access geo-restricted content or check localised pricing. |
| Rate limiting |
| When a website restricts how many requests a single IP address can make in a given time window. Proxies help by distributing requests across many IP addresses. |
| ASN (Autonomous System Number) |
| A number that identifies a network operator (e.g. Comcast, Amazon AWS, OVH). Websites use ASN to determine whether an IP comes from a consumer ISP (trusted) or a data centre (suspicious). |
| Success rate |
| The percentage of proxy requests that return a valid, unblocked response. The most important metric when evaluating a proxy provider. |
| CAPTCHA |
| A challenge-response test (like identifying traffic lights in images) used by websites to verify a visitor is human. A high CAPTCHA rate means a website’s bot detection has flagged your requests as automated. |
| ToS (Terms of Service) |
| The rules a website sets for using its service. Many sites prohibit automated access, scraping, or multiple accounts. Violating ToS can result in bans but is usually not a criminal offence on its own. |
