Shared proxy
A proxy whose IP address is used by more than one customer at the same time, cheaper than a dedicated address but with reputation you do not fully control.
A shared proxy splits one address between several customers at the same time. You pay less because you are not the only person using it, and for undemanding work that trade is perfectly rational.
The thing being shared is not bandwidth, it is reputation. An address is only as trusted as the sum of everything ever done through it, and on a shared IP that sum includes strangers you cannot see, audit or influence. If one of them hammers a target at three in the morning and the address lands on a blocklist, you arrive the next day to a block you did not cause and cannot explain.
The failure mode is worth picturing concretely, because it does not look like a failure. Your requests still connect. The proxy is up, the latency is normal, nothing errors. You simply start getting challenge pages, or empty results, or prices that do not match what a real visitor sees, on a target that worked yesterday. Shared-address problems present as inexplicable quality loss rather than as downtime, which is why they burn so much debugging time.
Free public proxies are this arrangement taken to its limit: unlimited undisclosed co-tenants on an address nobody is accountable for. Our own measurements make the shape of that clear. Of 589,918 free proxies we have discovered and tested, 362,018 never returned a single successful check, and the median lifespan of the ones that did work is about 145 hours. Sharing at that scale is why the numbers look like that.
So the rule is about what depends on the outcome. Sharing is a false economy exactly when a task needs a clean and predictable history, and a sensible saving when it does not.
Frequently asked questions
How many people share a shared proxy?
Providers rarely disclose it, and the number matters less than the fact that you cannot see, audit or influence any of them. Ask anyway, because a provider willing to state a figure is telling you something about how they operate. On free public proxies the answer is effectively unlimited.
How do I know if a shared IP is causing my problems?
Test the same request on a dedicated or fresh address. If it succeeds there and fails on the shared one, the address is your problem rather than your code. Shared-address failures typically present as inexplicable quality loss on a target that worked yesterday, not as downtime.
Are shared proxies ever the right choice?
Frequently. For undemanding work where a block costs you a retry rather than a result, you are paying less precisely because you are not the only user, and that is a rational trade. It becomes a false economy the moment a task depends on a clean, predictable history.
What is the risk of a shared proxy going bad?
You inherit consequences you did not cause and cannot explain. An address is only as trusted as everything ever done through it, so if a co-tenant triggers a blocklist entry overnight you arrive to a block with no visible reason. The failure is silent, which is what makes it expensive in debugging time.
Back to the full glossary.