Proxy pool
The full set of IP addresses a provider can route you through, and the number behind a rotating proxy.
A proxy pool is the set of addresses a provider can route you through. Buying access to a rotating proxy is not buying an IP, it is buying membership of a pool, and the gateway hands you a different member of it per request or per session. You never learn which one you are getting, and for most work you should not need to.
Pool size is the number the industry advertises loudest and the one that deserves the most scepticism, because two very different figures get quoted under the same word. The first is everything the provider has ever observed. The second is how many addresses are reachable, clean and in your target country at the moment you send a request. Only the second one has any bearing on whether your job succeeds, and it is always the smaller number.
We can put a hard figure on that gap, because we measure it on our own free pool and publish it. As of 11 August 2026 we had discovered and tested 589,918 free proxies. The number alive at that moment was 4,023. That is roughly one in 147. Of the rest, 362,018 had never returned a single successful check in their entire recorded history: not dead now, never alive at all. The median lifespan of the ones that did work was about 145 hours.
Free proxies are the harshest possible case and the ratio is not that brutal on paid residential supply. But the SHAPE of the problem is identical, and that is the transferable lesson. A headline built from cumulative discovery will always dwarf a headline built from live availability, and a provider quoting the first is not necessarily lying, only answering a question you did not ask.
So the useful questions are narrower than pool size. How many distinct addresses are live in the specific country you need. How is that measured, and how recently. What happens when the pool cannot fill your request, does it fall back to a neighbouring country silently or does it tell you. A provider that can answer those is describing a pool; a provider that can only quote a total is describing a marketing asset.
Size is not free of trade-offs either. A larger pool spreads traffic thinner and offers more geographic choice, which is genuinely valuable. It also means less control over what you get, and a pool assembled carelessly can be large and dirty at the same time, since every address arrives carrying whatever reputation its previous traffic earned it.
How HProxy handles it
We report what is live rather than a peak historical count, and the free-pool numbers above are published precisely because they are unflattering. 129,131,311 checks stand behind that 4,023 figure. The gap between a claimed pool and a working one is exactly where a proxy purchase goes wrong, and the only defence is measuring continuously rather than trusting a supplier's count.
Frequently asked questions
Does a bigger proxy pool mean better results?
Only if the extra addresses are alive, clean and in the country you need. Size helps by spreading your traffic so no single address accumulates enough volume to trip a limit, and by widening geographic choice. It does nothing if the count is cumulative rather than live, and a large pool assembled without care can be big and poorly reputed at the same time.
How do I verify a provider's pool size claim?
Ask a narrower question than they advertised: how many distinct addresses are live right now in the one country you actually need, how that is measured, and how recently. Then test it. Run a real job at your intended volume against your real target and count the distinct exit addresses you actually receive. The number you observe is the only one that has ever mattered.
Why do free proxy lists have so few working entries?
Because nobody is accountable for them. They are misconfigured or abandoned endpoints that anyone can find and everybody hammers, so they die quickly and are replaced by others that die just as fast. Our own measurement to 11 August 2026: of 589,918 discovered, 4,023 were alive, and 362,018 had never worked even once. Median lifespan for the ones that did work was around 145 hours.
What happens if the pool cannot fill my country request?
It depends on the provider, and it is worth establishing before you rely on it. Some return an error, which is honest and lets you handle it. Others silently fall back to a neighbouring country or a broader region, which is far worse for you, because your job appears to succeed while quietly collecting data from the wrong place. Test the failure case deliberately rather than discovering it in your results.
Back to the full glossary.