Fair Use Policies
This article explains how different proxy providers handle heavy usage. ProxyCompass information appears separately at the end.
A Fair Use Policy, or FUP, sets conditions on traffic volume and the load placed on a service. It is part of a provider’s service policies. It can explain why a paid package with unlimited traffic becomes slower, allows fewer simultaneous connections, or starts generating additional charges before its subscription ends.
Why do providers limit an unlimited service?
A proxy subscription gives you access to IP addresses and the equipment that serves them. Servers process connections, network links carry data, and the provider pays to keep both running. Server capacity and network speed are finite. Several customers’ proxies can run on the same equipment and use the same links.
Some customers run short jobs with long gaps between them; others transfer data continuously. If one customer occupies a large share of a network link for hours or places a heavy load on a server, other customers’ work can slow down. FUP defines the load a plan permits and when the provider may restrict it.
The number of IPs alone does not tell you how fast your work can run. A large list of addresses may be served by one server. Even a dedicated IP does not necessarily include a separate server or network connection. Buying more addresses gives you more IPs for reaching websites, but does not guarantee a higher total transfer speed.
This does not mean that all providers have a hidden traffic quota. Some sell unmetered access without a fixed monthly data cap. Others publish a volume threshold, and others reserve the right to intervene when sustained use affects the network. These are different ways of providing the service.
What exactly can be unlimited?
“Unlimited traffic” usually means data transfer without a fixed allowance measured in gigabytes. It does not mean that speed and simultaneous connections are unlimited too.
For example, a package can have no monthly traffic allowance but still have a maximum transfer speed. In a given time, you can only download as much data as that speed permits. The ceiling applies from the start. Transferring data at the maximum speed does not, by itself, mean that you have breached FUP.
Some plans reduce speed or the number of available connections after a traffic threshold. Access continues, but the job takes longer. Other plans include a set amount of data and charge separately for traffic beyond it. In the second case, work may continue at the same speed while the total cost rises.
The units help explain the limits. GB and TB describe an amount of data. Mbps and Gbps describe a transfer rate. A connection limit tells you how many connections can remain open at once. Sellers sometimes use bandwidth for both traffic volume and speed, so check the units used for the parameter.
For example, a program can download a large archive through one connection: there is only one connection, but a lot of data is transferred. Alternatively, it can open many connections and receive a short product-price response through each. Both jobs can use the same service. Few connections do not necessarily mean little traffic, and many connections do not necessarily transfer much data.
Why does the pattern of work matter?
A short busy interval and continuous heavy use place different demands on a service. A price-monitoring program might collect an update and then remain idle until its next run. Another program might keep downloading pages without a break. Even if both use the same type and number of IPs, their demand on shared capacity can be very different.
If a rule concerns traffic volume alone, the restriction takes effect when the stated daily or monthly threshold is reached. If the provider limits sustained heavy use, it may also consider how long the work continues and whether it affects other customers. In that case, low traffic consumption does not necessarily mean that continuous use at maximum speed or at the connection limit is permitted.
The product matters too. A mobile proxy relies on a cellular connection provided through a modem, router or phone. Heavy use can affect that connection and shared radio capacity as well as the proxy provider’s equipment. A policy written for mobile proxies cannot automatically be applied to a datacenter package.
There is no general rule that proxy automation or round-the-clock operation violates FUP. A service may be designed for sustained collection, or its terms may restrict that pattern. The relevant question is whether the product supports the workload you intend to keep running.
The purpose of the activity is a separate issue. An Acceptable Use Policy deals with permitted and prohibited tasks. FUP deals with resource use. A permitted collection task can create too much load for its plan; a prohibited task can break service rules while transferring very little data.
Why can traffic usage be high when there are few results?
Suppose a browser-based program saves only a product’s name and price. To obtain those two values, it may load the whole page, images, scripts and other resources. The saved table can be tiny compared with the data that travelled through the proxy. FUP accounting can concern that transferred data, not the useful records you kept.
Retries add another difference. A blocked page, an error response or a CAPTCHA can contain data even though the program obtains no useful result. If it keeps retrying, the traffic counter can grow while the results file hardly changes. On a byte-metered service, failed work should not be assumed to consume no allowance.
If the task does not need images or other page resources, avoiding those downloads reduces the data transferred. When requests keep failing, stop and establish the cause so the job does not keep consuming resources without a useful result. Traffic accounting and billing by successful results are separate conditions: a provider may charge only for successful requests while still applying limits to connections or other resource use.
Does the limit apply to each IP, a package or the whole account?
Imagine that most of the work goes through one IP while the other addresses in the package are idle. With separate per-IP allowances, that busy address can reach its limit even though the others have used little. Their unused allowances do not automatically become available to it.
If the allowance covers the whole package, traffic from all its IPs is counted together. One address can consume most of that total. Once it runs out, switching to another IP in the same package will not help: both addresses use the same allowance.
A common allowance can cover several packages in an account or all access to a particular proxy product. Replacing an IP, rotating addresses or creating another proxy login does not reset that common counter. Sharing an IP with other customers does not, by itself, mean they consume your traffic allowance. Rules about sharing addresses and rules for counting traffic are separate.
Some providers use two thresholds together: total traffic and usage per IP. A restriction may apply only when both are exceeded. Looking at just one can give a misleading estimate of the traffic available before a restriction. The same provider may also use different rules for shared and dedicated proxies.
What changes when FUP takes effect?
A restriction does not always make an IP unreachable. A proxy checker can still report that the addresses work while a production job becomes much slower. The consequence depends on which resource the provider changes.
| Provider action | Effect on your work |
|---|---|
| Lower transfer speed | Connections remain available, but pages and files take longer to transfer. Adding workers will not remove a speed ceiling that applies to the whole package. |
| Fewer simultaneous connections | Individual requests can still be fast, but fewer can run together. A job built around extensive parallel work can fall behind even though a simple connection test passes. |
| Charges beyond the allowance | The job continues and may finish normally, while additional usage increases the bill. A warning email does not necessarily stop the software or the charges. |
| Suspended access | Access stops for the affected proxies. Restoring it may require waiting for the allowance to renew, buying more traffic or addressing the cause of excessive load. |
These are alternatives used in different plans, not a compulsory sequence. Some providers apply a published restriction automatically. Others contact the customer about reducing the load or changing the service arrangement. A warning and time to adjust should not be assumed unless they are part of the stated process.
A slowdown alone does not prove that FUP has been applied. Ordinary network congestion, a normal plan limit and a destination website’s restrictions can produce similar symptoms. If a website blocks the IP, the provider may not have restricted the subscription at all. The distinction matters because buying more allowance will not necessarily solve a website-access problem.
What is needed to lift the restriction?
First separate current load from accumulated usage. Reducing active work can relieve the current load. It does not remove traffic already counted toward a daily or monthly allowance. An overnight pause can therefore help with one kind of problem and have no effect on the other.
If the traffic allowance has run out, the restriction may remain until the allowance renews or extra data is purchased. The renewal date depends on the plan: it may be the start of a calendar month or the next paid period. For a package bought halfway through a month, those dates differ.
An upgrade should address what your job actually needs. Extra data may help if the traffic allowance has run out. A higher connection limit may help if the program cannot open enough connections at once. More IPs help only if the relevant allowance increases with them. Simply adding money to the account balance does not necessarily change the package settings.
The lifetime of extra capacity also matters. An additional allowance may expire at the end of the current cycle, so a purchase near that date can have only a short useful life. A later upgrade should not be assumed to cancel extra usage charges already incurred.
For overage billing, a notification and a stop are different controls. A notification tells you about usage; a configured stop is intended to prevent further work or spending. If the service checks that stop periodically, some additional usage may occur before it takes effect.
If the provider suspended access because of excessive load, it may restore access only after you address the cause. Waiting for the next month will not necessarily help. Ask what needs to change in your program’s activity and when access will be restored. Refund eligibility is governed by the service’s separate refund rules.
How does this help you assess a plan?
Start with the work you need to complete: its schedule, expected data transfer and simultaneous connections. A package with many IPs but little parallel capacity may be unsuitable for a job with a tight completion window. A lower-speed unmetered plan may still suit a steady job without that deadline. An allowance-plus-charges model may preserve performance but make the total cost less predictable.
A short test can show whether the proxies connect and reach your websites. Running a typical job can also reveal how much traffic your software uses. But if you do not reach the FUP threshold during the test, it cannot show how the service will perform after a restriction. Initial speed and conditions for continued use need to be assessed separately.
Where a provider only says “excessive use” without giving a usable boundary, describe your expected workload and ask whether that specific plan supports it. An answer that merely repeats “unlimited” leaves the original uncertainty unresolved. You need to know what can change during the job, what would trigger that change and what continuing at the required capacity would involve.
The useful comparison is therefore how much work a plan can complete at an acceptable speed and total cost. IP count and the unlimited label are only part of that comparison.