Residential proxies route your requests through addresses belonging to real consumer internet connections, so the site you are talking to sees an ordinary broadband user rather than a server in a data centre. That is the whole product, and every provider in this article sells a version of it.
What they do not sell is the same price. Published rates for entry-level residential traffic currently span roughly $0.79 to $7.00 per gigabyte — close to a tenfold difference for something described identically on every website. We are Geonode, we sit at the bottom of that range, and we are one of the providers being compared here. Read accordingly.
The honest thing to do with that position is not to defend it but to explain why the low number is the wrong thing to lead with. A gigabyte that fails to get you the page is not cheaper than a gigabyte that does — it is infinitely more expensive, because you paid for it and got nothing. The metric that decides your actual bill is cost per successful request, and no provider can tell you theirs, because it depends on your target. There is a section below on how to measure it yourself in an afternoon, and it matters more than everything else in this article.
There are also two structural things that comparison tables almost never include and that regularly change the ranking: whether your unused traffic expires at the end of the month, and whether the price you are looking at is a promotional rate. One provider here advertises traffic that never expires; another displays a rate at 50% off with the regular price alongside it. Both facts move the real cost more than a few cents per gigabyte do.
Every figure below was taken from the provider's own live pricing page at the time of writing. They change frequently — sometimes weekly — so treat all of them as a starting point for your own check rather than a quote.
What Residential Proxies Actually Are
Worth being precise, because the category boundaries decide whether you should be reading this at all.
A residential proxy routes your request through a device connected via a consumer internet service provider. The address you appear from was allocated to a household broadband connection rather than to a hosting company. To the destination site, the request looks like it came from a person at home.
That matters because address origin is publicly checkable. Every IP block is registered to an organisation through the regional internet registries, and it takes seconds to determine whether an address belongs to a cloud provider or a consumer ISP. Sites that care about automated traffic check exactly this.
Where the Addresses Come From
Providers build these networks by obtaining access to real consumer connections — most commonly by compensating people who opt in through applications that share a portion of their bandwidth, or via software development kits embedded in apps with the user's consent.
This is worth understanding because it explains the price. Residential bandwidth carries a genuine ongoing acquisition cost per gigabyte, which datacenter bandwidth does not. It is also where the ethical questions in this industry live: how clearly consent was obtained, and how easily a participant can leave. It is a fair question to put to any provider, ours included, and a provider who cannot answer it clearly has told you something.
The Neighbouring Products
Datacenter proxies come from hosting providers. Five to ten times cheaper, considerably faster, and visibly datacenter addresses. Correct for any target that does not check — which is a great many of them.
ISP or static residential addresses are registered to consumer ISPs but hosted in data centres. They combine residential-looking origin with server-grade stability, are priced per address rather than per gigabyte, and are what you want for anything involving a persistent identity or a logged-in account.
Mobile proxies route through cellular carriers. The most expensive option, and worth it only where the target treats carrier addresses more leniently — which some platforms do and most do not.
Rotating residential, the subject of this article, is the right choice when you need many addresses that look like ordinary users, and you do not need any one of them to persist.
The Metric That Matters Is Not Price Per GB
This is the section that changes conclusions, so it comes before the table rather than after it.
Every comparison in this category ranks providers by dollars per gigabyte. That number is easy to obtain and easy to tabulate, and it is not the number that determines what you spend.
Why the Headline Rate Misleads
Consider two providers against the same target.
Provider A charges $1.00/GB and succeeds on 50% of requests. Provider B charges $2.50/GB and succeeds on 95%.
A looks 60% cheaper. Now count what you actually pay for. Every failed request still consumed bandwidth — the connection was made, the challenge page or the block response came back, and you were billed for it. To collect 1,000 successful pages you need roughly 2,000 attempts from A and roughly 1,053 from B.
If a page averages 500 KB, A costs about $1.00 per 1,000 successes and B about $1.32. A is still cheaper here, but the gap has collapsed from 60% to 24% — and that is before counting retry logic, engineering time, and the delay of running everything twice.
Push the success rate lower and the ranking flips outright. At 30% success, A needs 3,333 attempts and costs $1.67 per 1,000 successes against B's $1.32. The cheaper provider is now the more expensive one, and nothing on either pricing page would have told you.
The Uncomfortable Corollary
Success rates are target-specific and nobody can quote you yours. A provider that performs superbly against one marketplace may do poorly against another, and both may change next month when a target updates its detection.
This means published benchmarks — including any table — are directional at best. The only measurement that transfers is one you take yourself, against your own target, which is why the test described later matters more than everything else here.
What This Does Not Mean
It does not mean price is irrelevant. Once two providers both work on your target, the cheaper one is straightforwardly better and you should take it. The point is the order of operations: establish that it works, then compare price. Doing it the other way round is how people end up migrating three times.
The Providers Compared
Prices below come from each provider's own live pricing page at the time of writing. Verify before buying — several of these figures are explicitly promotional and this market reprices constantly.
| Provider | Entry rate | At volume | Free / trial |
|---|
| Geonode | $0.79/GB | $0.50/GB at 100 GB, $0.27/GB at 1 TB | 1 TB free for new accounts |
| Webshare | $3.50/GB (1 GB) | $1.50/GB at 1 TB, $1.40/GB at 3 TB | 10 proxies, 1 GB/mo, indefinitely |
| Decodo | $3.75/GB (3 GB) | $2.75/GB at 100 GB; $4.00 PAYG | 3-day trial, 100 MB |
| Bright Data | $4.00/GB PAYG (reg. $8) | $2.50/GB at 798 GB ($1,999/mo) | First deposit match to $500 |
| Oxylabs | $6.00/GB ($30 for 5 GB) | $2.50/GB at 1 TB ($2,500/mo) | — |
| IPRoyal | $7.00/GB (1 GB) | $1.75/GB from 10 TB | Traffic never expires |
Reading the Table
The entry rates span nearly 9x, which should immediately make you suspicious of the idea that these are the same product priced differently. They are differently positioned products, and the differences are in pool composition, geographic depth, success rate against hard targets, and the support you get when something breaks.
Bright Data and Oxylabs are the enterprise end, and their pricing reflects a different customer. Large pools, deep geo-targeting down to city and postcode level, compliance documentation, account management, and the sort of infrastructure that a Fortune 500 procurement department can be persuaded to sign off. At $6/GB entry, Oxylabs is not competing for the individual developer and does not pretend to be. If you need someone to answer the phone at 3 a.m. and produce a data processing agreement, this is where you look.
Bright Data's headline needs reading carefully. The $4.00/GB pay-as-you-go rate is displayed with "regularly $8/GB" beside it, and the subscription tiers likewise show 50% discounts. Whether that promotion persists into your renewal is a question worth asking before you build a budget on it.
Decodo and Webshare occupy the middle, at roughly $2.75–$3.75/GB in the ranges most people actually buy. Webshare's free tier is the most genuinely useful in the table — 10 proxies and 1 GB a month, indefinitely, with no card required. That is enough to run real tests, and enough to run a genuinely small job forever without paying.
IPRoyal has the highest entry rate and one structural advantage that we come to in the next section, which for lumpy workloads matters more than the rate.
Our own position is the lowest published entry rate here, and the section further down explains what that does and does not mean.
The Two Traps Nobody Tabulates
Both of these change the real cost more than a few cents per gigabyte, and neither appears in any comparison table we have seen.
Trap One: Traffic Expiry
Most residential proxy subscriptions expire monthly. You buy 25 GB, you use 8 GB, the other 17 GB vanish at the billing date. You have paid the advertised per-GB rate for gigabytes you never received.
This matters enormously for lumpy workloads, which is most real work. Scraping projects have bursts and lulls. A quarterly research job might use 40 GB in one week and nothing for two months. Under an expiring monthly plan, that pattern is brutal: you either buy a large plan and waste most of it every month, or buy a small one and run out mid-burst.
IPRoyal's residential pricing is explicit on this point — their FAQ states that the proxies "work for as long as you use up all the traffic", with traffic that does not expire. That is a genuine structural difference. At $7.00/GB it is the most expensive entry rate in the table, but if you use 30 GB across a whole year, non-expiring traffic at $7.00/GB costs $210 total, while a $3.00/GB monthly subscription sized to handle your peak week can easily cost more than that across twelve billing cycles of mostly-unused allowance.
Work out your usage shape before your usage volume. Steady monthly consumption favours subscriptions. Bursty or unpredictable consumption favours pay-as-you-go and non-expiring traffic, even at a higher headline rate.
Trap Two: The Price You See Is Not the Price You Renew At
Several providers in this market display heavily discounted rates as the default view. Bright Data's page shows 50% off with the regular rate beside it. Webshare's tiers show discounts of 50–80% against a $7.00 baseline.
Discounts are not dishonest — they are stated openly, and the discounted rate is a real rate. The problem is that a comparison table built from promotional numbers ranks providers by who is currently running the deepest promotion, which is not a durable property.
Before committing, ask two specific questions: what does this cost at renewal, and is this rate contractual or promotional. A provider that answers clearly is telling you something useful about how they will treat you later.
The Third Thing, Briefly
Check what counts as billable traffic. Failed requests, retries, and the overhead of connection establishment are usually billed. Some providers bill only successful requests on certain products. This can move effective costs by a substantial margin on a target with a poor success rate — which brings us back to the metric that actually matters.
What to Evaluate Beyond Price
Six things, roughly in the order they will cause you trouble.
Success rate on your target. Covered above, and it dominates everything else. Test it rather than reading about it.
Geographic depth where you need it. Every provider claims broad country coverage. Country-level is easy. What varies enormously is depth within a country — whether there are meaningfully many addresses in the specific city or region you need, or three addresses that all appear when you ask. If your work is geographic, test the exact locations rather than trusting a coverage map.
Session control. Can you hold the same address for a sequence of requests, and for how long? Rotating per request is right for wide crawls; sticky sessions are essential the moment a workflow spans several pages, because a login that changes country mid-checkout does not survive.
Concurrency limits. A generous monthly allowance is useless if you cannot spend it fast enough. Some providers cap concurrent connections by plan tier, sometimes without making it prominent. If your job has a time window, this is a harder constraint than volume.
Billing granularity and visibility. Can you see usage in near real time, set spend caps, and get alerts? A proxy bill can run away quietly, particularly when a retry loop meets a target that has started blocking. Spend limits are the cheapest insurance available.
Sourcing and compliance. How the network was built, what documentation exists, and whether the provider can support your own obligations if you are handling personal data. For an individual developer this is often a non-issue; for anyone operating under a compliance regime it is the first question, not the last.
What Not to Weigh Heavily
Advertised pool size. "Over 100 million IPs" is a number without a denominator. What matters is how many usable addresses exist in the places you need, at the times you work, that your target has not already blocked. A large tired pool is worse than a smaller clean one, and pool size is the easiest figure in the industry to inflate.
Published success-rate claims. Measured against targets the provider chose. Yours will differ.
Speed benchmarks. Residential proxies route through consumer connections; they are slower than datacenter and variable by nature. Unless latency is your binding constraint, success rate matters far more than milliseconds.
The Test That Decides It
An afternoon of this beats every comparison article, and it is not complicated.
Step One: Try Datacenter First
Before buying residential at all, run a few hundred requests against your real target through a datacenter proxy. If you are succeeding above roughly 95%, your target does not check address origin and you have just saved five to ten times your proxy budget. An enormous quantity of the web — documentation, public APIs, forums, small and mid-sized sites, most government and academic pages — does not check.
This step costs almost nothing and it is skipped constantly. Residential is the default purchase in this industry mostly because it is the default recommendation, not because it is usually necessary.
Step Two: Shortlist on Free Tiers
Every provider above offers a trial, a free tier, or a small entry plan. Webshare's indefinite 1 GB a month and our own 1 TB for new accounts both allow real testing rather than a token sample.
Pick two or three and set them up in parallel.
Step Three: Run the Same Job Through Each
Same target list, same volume, same concurrency, same time of day — that last one matters more than people expect, because residential pool composition shifts with when real users are online.
Record four numbers per provider:
- Successful responses, defined as containing the content you actually wanted rather than merely returning 200.
- Bandwidth consumed, including failures.
- Median and 95th-percentile latency.
- Geographic accuracy — spot-check that requests genuinely appear from where you asked.
Step Four: Divide
Bandwidth consumed × price per GB, divided by successful responses. That is your cost per successful request, and it is the only figure that compares providers meaningfully.
Run it and the ranking frequently differs from the price table. That is the point.
A Warning on Validating Success
Count successes by checking for content you expect, not by checking for HTTP 200. The characteristic failure mode here is a plausible page with the wrong contents — a challenge page, a soft block, a regional variant, an empty result set that looks like a legitimately empty result set. A success counter that trusts the status code will report excellent numbers while collecting nothing, and that error survives for a surprisingly long time.
Where We Fit and Where We Do Not
We are a provider in this comparison, so this section is the least neutral thing here. We have tried to make it useful anyway.
Our published rates are $0.79/GB entry, $0.50/GB from 100 GB and $0.27/GB from 1 TB, with 1 TB free for new accounts. Datacenter is $0.14/GB, billed by traffic rather than per IP, which is unusual in that category. ISP addresses are $1.25 each and unlimited residential is $1,800/month (current pricing).
That entry rate is the lowest in the table by a considerable margin, and the correct response to it is scepticism rather than enthusiasm. So here is what we would want to know in your position.
What a low per-GB rate does not tell you. It does not tell you the success rate against your target, which is the number that actually decides your cost. It does not tell you the geographic depth in the specific city you need. Both are measurable in an afternoon by you, and neither is measurable by reading anyone's pricing page — including ours.
Where we would expect to do well. High-volume work where bandwidth is the dominant cost and the targets are not the very hardest. Wide crawls, price and availability monitoring, market research at scale, work where the per-GB rate compounds into the whole budget. The 1 TB free tier exists so you can establish this without a purchase, and it is deliberately large enough to test properly rather than to produce a favourable demo.
Where we would not be the right choice. If you need enterprise compliance documentation, a named account manager and a procurement-friendly contract, the providers at the top of the price range are built for that and we are not going to win that comparison on paper. If your target is one of the hardest — the platforms that invest most heavily in detection — test carefully rather than assuming a rate advantage translates. And if your requirement is a persistent identity for account work, rotating residential is the wrong product entirely; you want static ISP addresses, from us or from anyone.
The recommendation we would give against ourselves. If your volume is genuinely small — a gigabyte or two a month — Webshare's indefinite free tier will cost you nothing at all, and "nothing" beats $0.79. If your usage is lumpy and annual rather than monthly, IPRoyal's non-expiring traffic may cost less in total than any monthly plan including ours, despite the higher headline rate. Both of those are real, and we would rather you knew.
When You Should Not Buy Residential
Against our interest, and the section we would most like people to read.
Your target does not check address origin. The single most expensive mistake in this market. Test with datacenter first — at $0.14/GB against $0.79/GB and up, the saving is most of an order of magnitude, and a large share of the web genuinely does not care.
You are making a handful of requests. Rate limiting and detection respond to volume and pattern. Fifty requests spread across a day look like a person. Buying proxies for a small job solves a problem you do not have.
You control the target. Load-testing your own service, monitoring your own endpoints, checking your own site. Route directly and measure reality rather than reality plus a proxy hop.
You need a persistent identity. Account management, anything with a login, anything where the same address should appear every day. Rotating residential is actively wrong here — an account that appears from a new city each session looks worse than one that never moved. Static ISP addresses are the product.
The problem is not your address. Missing headers, absent cookies, a fingerprint that gives away automation, a request rate no human would produce. All of these produce blocks that look identical to address-based blocks, and none of them are fixed by better proxies. Diagnose first; buying infrastructure to work around a malformed request is an expensive way to keep the bug.
The page is rendered by JavaScript. No proxy makes an empty shell into content. You need a headless browser, and swapping providers while the page stays blank is a well-worn afternoon.
The data is available another way. Official APIs, public datasets, bulk downloads, an RSS feed. Scraping is the fallback when there is no supported route, and checking for one takes ten minutes.
You have not read the terms. Automated collection is frequently against a site's terms of service, and personal data carries obligations regardless of how it was collected. Proxies change what a site can observe; they do not change what you agreed to or what the law requires. If this matters in your context, get advice rather than a blog post.
People Also Ask
What is the best residential proxy provider?
There is no single answer, because the deciding factor is success rate against your specific target and nobody can quote you that. Bright Data and Oxylabs lead on enterprise features and compliance at $4–$6/GB entry; Decodo and Webshare sit in the middle around $2.75–$3.75/GB; Geonode publishes the lowest entry rate at $0.79/GB. Test two or three on free tiers against your own target and compare cost per successful request.
How much do residential proxies cost?
Published entry rates currently run from about $0.79 to $7.00 per GB, dropping substantially at volume — to roughly $1.40–$2.50/GB at terabyte scale across most providers, and lower on some. Several displayed rates are promotional, so check what the renewal price will be.
Is a cheaper proxy provider worse?
Not necessarily, but the price alone cannot tell you. What matters is cost per successful request: a $1/GB provider succeeding half the time can be more expensive in practice than a $2.50/GB provider succeeding almost always, because failed requests consume paid bandwidth. Measure it on your target rather than inferring it from the rate.
Do I need residential proxies or will datacenter do?
Run a few hundred requests against your real target through datacenter first. Above roughly 95% success, datacenter works and you have saved most of an order of magnitude. Blocks, CAPTCHAs or empty responses mean the target checks address origin and you need residential.
Does unused proxy traffic expire?
Usually yes — most monthly subscriptions expire unused gigabytes at the billing date. IPRoyal is a notable exception, stating that traffic does not expire. For bursty or seasonal workloads this single term can outweigh a substantial difference in headline rate.
Is there a free residential proxy?
Webshare offers 10 proxies with up to 1 GB a month indefinitely with no card required, which is the most useful ongoing free tier. Geonode gives new accounts 1 TB, Decodo a 3-day 100 MB trial, and Bright Data matches a first deposit up to $500. Avoid free public proxy lists entirely — the operator sees all your traffic and has no reason to be trustworthy.
Are residential proxies legal?
Using a proxy is lawful in most jurisdictions. What you do through one is governed by the same laws and contracts as anything else — site terms of service still apply, and collecting personal data still requires a lawful basis. This is general information rather than legal advice.
How many proxies do I need?
With rotating residential you are buying bandwidth rather than a count of addresses, so the question is how many gigabytes your job consumes. Estimate average page size multiplied by pages, then add a generous margin for failures and retries — which on a difficult target can double it.
Wrapping Up
The residential proxy market publishes entry rates spanning nearly nine times for a product described identically everywhere, which is a reliable sign that the headline number is not measuring the thing you care about.
The thing you care about is cost per successful request. Failed attempts consume paid bandwidth exactly like successful ones, so a low rate against a target where you succeed half the time is not the bargain it appears to be, and the ranking genuinely inverts at sufficiently poor success rates. No provider can quote you this figure, because it depends entirely on what you are fetching. You can measure it in an afternoon, and that measurement is worth more than every comparison table published on the subject.
Two structural terms deserve checking before the rate does. Does unused traffic expire, which for bursty workloads can matter more than several dollars a gigabyte — IPRoyal's non-expiring traffic at $7.00/GB genuinely beats a cheaper expiring subscription for annual or seasonal use. And is the displayed price promotional, because a table built from temporary discounts ranks providers by who is running the deepest sale this week.
Before any of it, spend ten minutes finding out whether you need residential at all. Datacenter at $0.14/GB works on a large share of the web, and testing it first is the single highest-return action in this entire process.
We publish the lowest entry rate in the comparison and we are still telling you to test datacenter first, to check Webshare's free tier if your volume is tiny, and to look at non-expiring traffic if your usage is lumpy. That is not modesty — it is that a customer who bought the wrong product costs everyone more than the sale was worth.
Run the test. Divide bandwidth by successes. Then buy the cheapest thing that actually worked.
