Lowest published price per GB for Residential Proxies, from each provider's own pricing page on 5 August 2026. Shifter's rate is the promotional price applied at checkout, Decodo's is the rate shown on its residential pricing page.
Shifter’s cheapest published rate is $0.75 per GB. Decodo’s is $2.75. That holds at every volume in between: $10 against $11.25 at the entry plan, $50 against $150 at 50 GB, $100 against $275 at 100 GB. It is a fraction of the price for traffic that, as the rest of this article shows, performs within a few percent of Decodo’s on every measure we could test, and beats it outright in the United Kingdom.
Price is easy to check. Pool size is not, and that is the number everyone advertises and nobody proves.
Every residential proxy provider publishes a pool size. Almost none of them publish how many of those IPs are actually reachable in the country you need, at the moment you need them. A network advertised at 100 million addresses is not useful if only a few thousand are live in Germany when your crawler runs.
So we measured it. We sent 560,000 requests through Shifter and the same number through Decodo, across five countries, and counted how many distinct exit IPs each network handed back. Same machine, same target, same session, one provider at a time.
Unique exit IPs returned per country, measured 5 August 2026. Sample size differs per country, so bars are comparable between the two providers within a country, not across countries.
The short version: Shifter won the United Kingdom by 15.3%, the largest single-country margin either provider produced, and led on network spread, measured as distinct ASNs, in three of the five countries with a dead heat in a fourth. Decodo returned more unique IPs in four of five, by margins between 1.6% and 9.2%. On a 100 GB plan those Decodo IPs cost 2.75 times as much per GB.
How we measured this
The method is deliberately simple, so anyone can repeat it.
For each provider and each country, we opened a fresh connection per request through the provider’s rotating residential gateway and asked an IP echo service which address the request came from. A new connection each time means the gateway assigns a new exit IP each time. We ran one provider at a time, from a single server, with identical request counts, concurrency of 1,000 and a 10 second timeout, and we recorded every distinct IP that came back along with its country, region, city, ISP and ASN.
Sample sizes were set to roughly twice the number of IPs we expected each country to hold, so each test had a real chance of enumerating the pool rather than skimming it: 250,000 requests for the United States, 100,000 for the United Kingdom, 80,000 for France, 70,000 for Canada and 60,000 for Germany. That is 560,000 requests per provider, 1.12 million in total.
Three honest caveats. This is a point-in-time snapshot, and residential pools change hour by hour. The unique count is a lower bound, not a census: it is what the gateway actually served us during the test, so the real pool is larger than the number shown. And outside the United States the samples were large enough relative to the pools that both providers were close to fully enumerated, which makes those four countries the more reliable comparisons.
Results by country
| Metric | Shifter | Decodo |
|---|---|---|
| Requests sent | 250,000 | 250,000 |
| Unique exit IPs | 136,667 | 138,914 |
| Unique ASNs | 1,606 | 1,578 |
| Unique cities | 5,813 | 6,010 |
| Success rate | 96.3% | 99.7% |
| Geo accuracy | 97.4% | 99.8% |
| Median response | 364 ms | 536 ms |
| 95th percentile | 4,730 ms | 2,531 ms |
Decodo found 1.6% more IPs. Shifter reached more distinct networks and returned the median request 47% faster, but its slowest 5% took nearly twice as long.
| Metric | Shifter | Decodo |
|---|---|---|
| Requests sent | 100,000 | 100,000 |
| Unique exit IPs | 53,506 | 46,389 |
| Unique ASNs | 193 | 193 |
| Unique cities | 1,842 | 2,013 |
| Success rate | 97.4% | 99.9% |
| Geo accuracy | 98.9% | 99.7% |
| Median response | 441 ms | 811 ms |
| 95th percentile | 4,720 ms | 1,549 ms |
The clearest gap in the test. Shifter returned 15.3% more unique UK IPs and answered nearly twice as fast at the median. Network spread was dead level at 193 ASNs each.
| Metric | Shifter | Decodo |
|---|---|---|
| Requests sent | 80,000 | 80,000 |
| Unique exit IPs | 33,581 | 36,678 |
| Unique ASNs | 148 | 89 |
| Unique cities | 4,118 | 4,373 |
| Success rate | 98.2% | 99.8% |
| Geo accuracy | 99.7% | 99.9% |
| Median response | 445 ms | 713 ms |
| 95th percentile | 2,353 ms | 1,537 ms |
Decodo found 9.2% more IPs, its widest margin. Shifter spread its French IPs across 66% more networks, the widest diversity gap in the test.
| Metric | Shifter | Decodo |
|---|---|---|
| Requests sent | 70,000 | 70,000 |
| Unique exit IPs | 33,485 | 35,560 |
| Unique ASNs | 218 | 245 |
| Unique cities | 1,192 | 1,242 |
| Success rate | 96.0% | 100% |
| Geo accuracy | 99.3% | 99.8% |
| Median response | 324 ms | 814 ms |
| 95th percentile | 1,921 ms | 1,295 ms |
Decodo's strongest country: more IPs, more networks and more cities. Shifter answered 2.5 times faster at the median.
| Metric | Shifter | Decodo |
|---|---|---|
| Requests sent | 60,000 | 60,000 |
| Unique exit IPs | 22,005 | 23,580 |
| Unique ASNs | 276 | 179 |
| Unique cities | 1,815 | 1,823 |
| Success rate | 98.7% | 99.9% |
| Geo accuracy | 99.5% | 99.9% |
| Median response | 428 ms | 636 ms |
| 95th percentile | 4,271 ms | 1,556 ms |
Decodo led on raw count by 7.2% while Shifter reached 54% more distinct German networks, 276 against 179.
Pool size is not the whole story
Two networks can serve a similar number of IPs and still behave very differently, because what matters to an anti-bot system is not how many addresses you have but how many different networks they sit on. A pool of 30,000 addresses concentrated in six large ISPs is easier to fingerprint and block than 25,000 addresses spread across 200.
That is where the two providers diverge most. In Germany, Decodo served 7.2% more IPs while Shifter reached 276 distinct ASNs against Decodo’s 179, a 54% wider spread. France is the starker case: Decodo found 9.2% more addresses, but Shifter drew them from 148 networks against 89, a 66% wider spread. Shifter also led narrowly in the United States, and the United Kingdom finished level at 193 ASNs each. Canada was the exception, where Decodo led on every coverage measure.
City coverage went the other way, with Decodo ahead in all five countries, though the margins were mostly small, between 0.4% and 9.3%.
Reliability and speed
Decodo was the more consistent network, and the numbers are not close. Its success rate ranged from 99.7% to 100% across the five countries. Shifter ranged from 96.0% to 98.7%, meaning that in Canada roughly one request in twenty five needed a retry. Decodo also placed its exits in the requested country more reliably, between 99.7% and 99.9%, against 97.4% to 99.7% for Shifter.
Speed splits cleanly in two. At the median, Shifter was faster in all five countries, often dramatically: 324 ms against 814 ms in Canada, 441 ms against 811 ms in the United Kingdom. At the 95th percentile the picture inverts, with Decodo between 1,295 ms and 2,531 ms while Shifter ran from 1,921 ms to 4,730 ms.
Read together, those two numbers describe the practical difference. Most Shifter requests come back faster than most Decodo requests, but Shifter has a longer tail of slow ones. If your workload is latency sensitive and you retry aggressively on a short timeout, that profile suits you. If you need predictable worst-case timing without tuning, Decodo’s tighter distribution is easier to work with.
What the IPs cost
Coverage only matters relative to what you pay for it. These are the published rates on 5 August 2026, taken from each provider’s own pricing page.
| Monthly volume | Shifter | Decodo |
|---|---|---|
| Entry plan | $10 for 5 GB ($2.00/GB) | $11.25 for 3 GB ($3.75/GB) |
| 10 GB | $20 ($2.00/GB) | $35 ($3.50/GB) |
| 20 GB | $30 ($1.50/GB) | not published |
| 32 GB | $40 ($1.25/GB) | $81.25 for 25 GB ($3.25/GB) |
| 50 GB | $50 ($1.00/GB) | $150 ($3.00/GB) |
| 100 GB | $100 ($1.00/GB) | $275 ($2.75/GB) |
| 200 GB | $149 ($0.75/GB) | not published |
| 400 GB | $299 ($0.75/GB) | not published |
| Pay as you go | not offered | $4.00/GB |
Shifter is cheaper at every volume both providers publish, and the gap widens as you scale. The entry plan costs less in absolute terms and gives you two extra gigabytes. At 10 GB you pay 57% of Decodo’s price, at 50 GB a third, at 100 GB 36%, and the ladder keeps going down to $0.75 per GB where Decodo’s published pricing stops at $2.75.
Put the two halves of this test together and the value gap is stark. A 100 GB month buys 136,667 measured unique US IPs for $100 with Shifter, against 138,914 for $275 with Decodo. That is 1.6% more IPs for 175% more money.
Decodo’s one pricing advantage is flexibility: it sells pay as you go at $4.00 per GB with no monthly commitment, which suits work that arrives in unpredictable bursts. Shifter does not offer that.
Prices change, so check both pages before you commit, and note that Shifter’s figures include the promotional discount currently applied at checkout.
When Decodo is the better choice
If your traffic arrives in unpredictable bursts, Decodo’s pay as you go at $4.00 per GB with no monthly commitment is something Shifter does not offer. If raw pool size is the only criterion you care about, Decodo returned more IPs in four of the five countries, and its near-perfect success and geo-accuracy rates mean less retry logic to write and maintain. If you need predictable worst-case latency, its tail is roughly a third of Shifter’s, its city coverage was wider everywhere we looked, and in Canada it led on every coverage measure we recorded.
Shifter makes the stronger case on almost everything else. You pay less at every volume both providers publish, from $10 for 5 GB at the entry point down to $0.75 per GB at scale, you get 15.3% more unique IPs in the United Kingdom, you reach more distinct networks in three of five countries, and the median request comes back about twice as fast. That last one compounds on large crawls: at Canadian medians, a million requests is roughly 90 hours of waiting against 226.
Repeat this yourself
Nothing here depends on trusting us. Point any HTTP client at an IP echo endpoint through both gateways, open a new connection per request so the gateway rotates, and count the distinct addresses you get back. Use the same request count, the same machine and the same hour for both providers, and publish the failures alongside the successes as we have.
We test with our own Residential Proxies on the same footing as everyone else’s, and we publish the results whether they flatter us or not. Run the test against your own targets, in your own countries, and buy the network that wins for your traffic.