Proxy Market Research 2022
For four years running, Proxy Market Research remains the largest – and only! – in-depth evaluation of the leading proxy services. Though focused on technical benchmarks, the research gives a complete overview of the providers, along with a glimpse into the broader state of the market. It can benefit anyone using or considering a proxy service, whether you’re an individual in need of proxies, a business, or a proxy seller looking to improve your products.
This year’s research covers 11 providers. These include companies like Bright Data, Oxylabs, and Smartproxy, which constitute the lion’s share of the market, together with smaller services like IPRoyal and PacketStream. We aim to make as broad a representation as possible while keeping the scale manageable.
Compared to the previous year, the 2022 edition makes a few changes. First, it benchmarks more products: not only residential, but also dedicated datacenter and mobile proxies – even several web scrapers. Second, it puts greater emphasis on individual country pools to more accurately evaluate the proxy networks. And finally, it comes as a web page (and not a .pdf!) to make the content more interactive. For those who want it, a .pdf version will be available at a later date.
- Datacenter and residential proxies remain the most popular proxy types, with 95% of use. But proxy-based APIs are quickly picking up the pace.
- Businesses mostly buy proxies to collect data from e-commerce and travel websites, monitor their marketing performance on search engines, and help with their ad efforts.
- Last year brought many small mobile proxy shops, while sneaker scalping experienced a major hit due to fewer releases and tightening security.
- Providers released five new bandwidth sharing apps in effort to source IPs more transparently.
- Providers are still fighting in court over the right to provide residential proxy services.
- You can find the evaluation graph here.
Dedicated Datacenter Proxies
- Most datacenter proxy services charge for IPs, while giving unlimited traffic and connection requests. Bright Data uses the most imaginative pricing model, which also includes the number of available domains, gigabytes, and even ability to target cities.
- Some providers demand up to 60% more for proxies outside of the US.
- Datacenter IPs are fast: they reached popular targets 2-3 times faster than residential addresses (1.75 response time on average). However, they also got blocked more, especially on targets like Amazon and Craigslist.
- IPRoyal’s proxies were the fastest, while Blazing SEO and Oxylabs displayed the best overall success rates.
- All in all, Blazing SEO has the most balanced service, while Oxylabs should appeal to premium customers. You can find the evaluation graph here.
- Residential proxy services can be pretty clearly segmented into cheap, mid-level, and premium. The cheapest providers cost up to 10 times less than the premium ones.
- Nearly all providers are able to connect with over 90% success rate. Real performance differences manifest in response time, which depends on IP location. The fastest providers are three times faster than the slowest ones.
- Pool size is another differentiating criterion that very much favors the premium providers. Bright Data, Oxylabs, and Smartproxy had IP pools that were both larger and better balanced than most competitors.
- Nowadays it’s easy to find companies with over 100 locations, city targeting, and flexible rotation options. ASN targeting, however, remains a rare premium feature, and more than half of the participants fail to support SOCKS5.
- Residential proxy services still run on traffic-based subscriptions, and paying as you go is relatively rare.
- Overall, Bright Data, Oxylabs, and Smartproxy remain the best residential proxy providers. You can find the evaluation graphs here.
- Peer-to-peer mobile proxy services are few and far between compared to services that use dedicated devices.
- They’re significantly more expensive than residential IPs ($42.50 vs $13.64 at 5 GBs of data) but the differences between providers are not as pronounced. The dominant pricing model is traffic-based subscription.
- With one exception, the proxy networks have nearly perfect success rates but connect half as fast as residential addresses. Oxylabs is the exception: its proxies outperformed 7 residential (!) proxy services.
- Mobile proxy pools are generally smaller. But the marketing numbers can’t always be trusted: SOAX had more unique IPs than Bright Data despite having half the advertised pool. Oxylabs won the comparison by a large margin.
- Blazing SEO uses a unique model that constantly rotates dedicated devices. It performed poorly due to being a beta service but generated a large number of IPs over time with a limited number of devices. This shows interesting opportunities for web scraping tasks that value IP uniqueness over speed.
- Overall, Oxylabs convincingly displayed the best results in most categories but price. You can find the evaluation graphs here.
- Proxy-based APIs add web scraping logic onto proxy networks to retrieve data with 100% success. Providers like Bright Data and Oxylabs have made them the only way to access certain protected websites.
- We based our tests on Google Search and chose the APIs accordingly.
- Such services charge for successfully retrieved data; at $100, they cost around $3.18 per 1,000 requests. Blazing SEO is the most affordable option.
- Bright Data’s tool showed impressive performance: with an average response time of 3.92 seconds, it outpaced the competitors by two to four times.
- Three out of the four participants could parse most aspects of Google Search, two were able to function asynchronously, and only one had CSV support.
- Overall, Bright Data and Oxylabs have the best proxy-based APIs for scraping Google. You can find the evaluation graphs here.
- The majority of the providers offer self-service, and two thirds have implemented a wallet functionality to reduce needless transactions.
- Most have straightforward proxy setup procedures but skimp on data visualization. NetNut and Bright Data have implemented the most detailed usage graphs.
- Providers tend to restrict their account management APIs for regular users, limiting its accessibility to resellers.
- Bright Data, Oxylabs, Smartproxy, and GeoSurf have the most detailed documentation about their services. It includes integration instructions, FAQs, code samples, and even white papers.
- Support by chat is much faster than over email, but only seven out of the 11 providers support it. The difference between the fastest and slowest average response exceeded 12 hours.
- Overall, we consider Smartproxy to have the best overall user experience. Bright Data left us with a good impression, too. You can find the evaluation graphs here.
A proxy generally means a substitute. It’s a server standing between two devices on a network that captures and routes connection requests through itself.
This research concerns proxy servers that are forward and highly anonymous. Forward means that the proxy server reroutes connection requests leaving a device, as opposed to reverse proxies that intercept requests reaching a website. The latter are used for a whole different set of tasks like load balancing. Highly anonymous refers to servers that hide the IP address and location of the user, disclosing only the identifying information of the proxy server.
Such proxy servers are most frequently categorized by their origin. The four main types are datacenter, residential, ISP, and mobile proxies. Lately, it also makes increasing sense to include proxy-based APIs which build upon individual proxy networks.
From improving anonymity to collecting huge amounts of data, proxy servers have many uses. We don’t pretend to know all of them – especially all the black-hat schemes that still haunt the web. Nonetheless, we decided to list some of the main business applications, together with the proxy types we believe work well for them. As always, advice like this should be treated as guidelines and might not always hold water when faced with individual circumstances.
You can switch to different tabs by selecting them.
Let’s review where the market headed in 2021. This year, we identified four noteworthy developments. The section also includes some first-hand proxy server usage trends. The latter are possible thanks to our partnership with Bright Data, which generously provided the information and allowed us to make it public. We’d like to thank the provider for that.
Proxy Server Usage Trends (in Partnership with Bright Data)
- Datacenter and residential addresses remain by far the most popular proxy types. Whether in the form of proxy networks or Web Unlocker, a proxy-based API, they take 95% of proxy usage. Despite the buzz, mobile and ISP proxies are still used for a narrow range of tasks, at least among Bright Data’s clientele.
- The proxy server market continues grow fast. The use of datacenter traffic increased by 60%, and residential proxies grew by 70% in 2021. It shows that more companies are finding value in collecting data and using it to make informed decisions.
- Proxy-based APIs are starting to play an increasingly important role. The signs were already there last year, but it seems that proxy-based APIs are now gaining real momentum:
- Web Unlocker increased its revenue by 60% by only running 10% more traffic. It shows that Bright Data and other providers are able to continuously optimize the tools through machine learning and basic trial-and-error.
- Search Engine Crawler, an API for scraping search engines, increased its traffic by 800%.
- Four major use cases constitute two thirds of all proxy use:
Explosion of Dongle-Based Mobile Proxy Services
Dongle-based mobile proxies are USB sticks with SIM cards inside them. With the help of special software (or a clever script), each stick turns into its own proxy server. Compared to peer-to-peer proxy networks, this arrangement gives more control over individual IPs, with longer uptime and on-demand rotation. And most importantly – it removes traffic limits, thanks to unlimited plans offered by mobile carriers.
Dongle-based mobile proxies aren’t new – the oldest services date back to 2018. However, it wasn’t until 2021 that this type of service really exploded. A quick scroll through BlackHatWorld, a wildly popular online forum, shows nearly 20 new providers that entered the market in 2021.
This number is all the more staggering considering that few of these services resell proxies (at least we’re inclined to think so). The whole gist is that you can set up the mobile proxies at home; the process is well documented, and even one device can potentially cover hundreds of IPs over time. A company called Proxidize further simplifies things, offering a plug-and-play solution that doesn’t require coding skills. According to the website, it’s served over 1,000 clients since launch.
The demand is driven by multiple account managers that need stable high-quality IPs for social media and e-commerce platforms. Some of the providers we spoke with have turned their hustles into million dollar businesses. But at the same time, out of the 20 newcomers, 5 have already closed. The proxy market juggernauts aren’t too keen (or hurried) on introducing dongle-based proxies either, at least so far.
Hard Times for the Sneaker Scalping Business
The last few years were generous for sneaker scalpers: money good and morale high. Minor setbacks – such as Nike’s SNKRS app becoming realistically unbottable – were offset by plentiful bounties from Yeezy Supply and Footsites. Even Supreme didn’t take security seriously yet.
In the process, sneakerheads popularized the category of ISP proxies by adapting its strengths for the niche. At first, this proxy type emulated the business model of residential proxies: it sold access to a pool IPs limited by traffic. Botters introduced IP-based plans without traffic limits, which displaced increasingly ineffective datacenter proxies.
The high continued well into 2021, reaching its apex in late spring. But then it took on a downward course. Without prying into internal data of proxy providers, we have two sources to tell the story: bot prices and Google search demand. Both show a sharp decrease starting with mid-summer:
What happened? An interview with a sneaker reseller breaks down the reasons well. First, Yeezy Supply started running dry with increasingly few coveted releases. Second, Foot Locker transitioned to online raffles, which broke major bots and made mass-purchases complicated. The third mentioned reason is dubious: US businesses like StockX and Venmo now have to report sales over $600 to the IRS, which makes it harder for small-time hustlers to avoid taxes.
By the end of 2021, many sneakerheads had turned away from reselling shoes in favor of more profitable avenues: consoles, GPUs, and NFTs. They’re still more desirable, but not without questions of their own (be it stock levels for consoles, Ethereum’s transition to proof of stake for GPUs, or simply hype in the case of NFTs). Sneakers should rebound once the bots adjust and drops return; but the glory days are probably over, and part of the community will stick with NFTs and other cash-cows while they’re so profitable.
A Proxyware App for Everyone
Proxyware bandwidth sharing apps offer people to rent out their IP in exchange for money. This way, they effectively become proxies, bolstering the networks of residential and mobile proxy providers. For each gigabyte of data transferred via their network, users earn up to a dollar (usually $.10-30), which they can spend on anything they desire, such as paying the Netflix bill.
Proxyware apps are a sought-after way to source IPs. While not without blemishes, this method ensures both explicit consent and compensation for the user. Proxy providers can then brag about their ethicality, which is becoming a serious selling point. They also decrease reliance on shady in-app SDKs that tend to get purged from mobile stores often enough to cause a headache.
Before 2021, the market had two major proxyware apps: Honeygain and PacketStream. At the end of the year, the number had more than tripled, with at least seven apps now on the market. Most of them are operated by major proxy providers wishing to jump onto the passive income train. This means more competition, more IP overlap (people have several apps installed at once), and hopefully, better conditions for the sharers.
The Legal Battles Continue
The proxy server industry is still in legal turmoil, with Bright Data trying to enforce patents against its biggest competitors. We know of three ongoing disputes; two carry on from earlier, while the third started mid 2021.
The first conflict involves Bright Data and GeoSurf. It began in late 2018 over alleged illegal use of trade secrets and patent infringement. Publicly, the lawsuit culminated in mid-2021 with a settlement. Bright Data promptly issued a public release that GeoSurf would give up proxy services and transfer its clients over. So far, this hasn’t happened. We asked GeoSurf for comment and were told that the US court has stopped any enforcement activity and GeoSurf is free to continue operating in full (while moving forward with its appeal process).
The second conflict takes place between Bright Data and Oxylabs, once again over alleged patent infringement. The dispute is in its fourth year now, with three out of five cases (so far) open. In 2021, the parties managed to resolve one case, and a jury awarded Bright Data $7.5 million in another. Oxylabs stoked the fire by filing a retaliatory lawsuit of its own in early 2022… again over patents. It looks like the dispute won’t end for a while (you can find its full timeline here). Until then, both parties conduct business as usual.
The third conflict began in the summer of 2020, between Bright Data and NetNut. As is tradition, the former sued its competitor over patents, more specifically, the datacenter proxy technology. The case ended in late 2021 by settlement, but there are more in progress: one initiated by Bright Data over residential proxies in summer 2021, and one filed by NetNut this March.
To reiterate our question from last year: What’s at stake? The outcome of these cases will decide whether Bright Data will have the power to quash any competitor that threatens its leadership with similar services. It may choose not to do so, but the mere ability will have chilling effects if it materializes. Even now, providers like SOAX deliberately exclude proxies in Texas where the litigation takes place to avoid getting entangled.
This year’s research includes 11 proxy services. Together they take the lion’s share of the proxy server market – especially in the areas of residential proxies and proxy-based web scraping APIs.
Many of these companies have appeared in our previous researches, so it’s possible to track their progress over time. Some, like RSocks and Shifter, rejoin after a year’s pause. Others – more specifically, Blazing SEO and IPRoyal – participate for the first time.
We first reached out to providers asking if they wanted to take part. Out of the 11 participants, 10 voluntarily gave access to their services; PacketStream failed to respond, so we used the funds we had on the platform from previous tests.
To evaluate the providers, we applied both automated performance tests and manual analysis.
The testing took place during February and March of 2022.
This year’s research has a bigger scope compared to previous editions. It includes not only residential proxies but other services as well: dedicated datacenter, pool-based mobile proxies, and proxy-based APIs.
|Participants||Dedicated Datacenter||Residential||Mobile||Proxy-Based API|
We ran automated performance tests using a custom web scraping script. It makes connection requests to specified websites and stores the results in a database. The server that made the requests was located in Germany.
The main target for proxy pool tests was Cloudflare’s CDN-CGI Trace tool which pings the closest available Cloudflare server. The website doesn’t block IPs and is always available, letting us get accurate information about a proxy provider’s infrastructure. Benchmarks with popular data collection targets used a modified version of the same script outfitted with user agents and other web scraping logic.
We enriched the output with data from MaxMind and other IP databases. This added data points like IP location, ASN, type, and whether the IPs are known as proxies.
To test the connection speed of proxies, we ran them through the DigitalOcean speed test.
The manual analysis involved collecting information about the providers’ features, user experience, pricing, and other details by hand.
To evaluate the providers, we distinguished categories that we deemed relevant for that proxy type, such as proxy pool, performance, or price. We then identified aspects within those categories – for example, number of unique IPs under pool size – and weighted them. Finally, we gave providers in each category a score from 0 to 5 based on their results.
The purpose of the evaluation methodology is to allow a quick comparison of the providers. We believe that the scores accomplish this task well. Still, some find the evaluation opinionated or the criteria too rigid (a pitfall of most standardized comparisons). In such cases, there’s always hard data to fall back on.
Here’s our evaluation of the participants based on the different services they provide. For your convenience, we’ve divided them into categories by price.
Approachable price, reliable performance, and availability typically make datacenter proxies the first choice for proxy server users. They can be shared among several people, and it even makes sense to talk about rotating datacenter proxy pools. But in general, businesses prefer addresses dedicated to one user, as it ensures full control over the IPs.
It’s both easy and hard to test datacenter proxies. Easy because you get a finite number of addresses – they either work with websites or don’t. Hard because the same set of IPs may not perform the same on different websites, and some providers, especially premium ones, hand-pick addresses for particular uses.
Still, that didn’t stop us from trying out dedicated datacenter proxy services. You’ll find this section divided into four parts:
- Proxy performance, which measures the success rate and detection status of each provider’s proxies.
- Features, which compares the functionality of the proxy services,
- Price, which investigates how much the services cost, and
- Wrap-up, where we briefly summarize the data and put the first three aspects into neat graphs. If that’s all you need for now, click here.
Dedicated datacenter proxies have the best raw specifications out of all proxy types: fastest connection, highest stability, and best performance under load per IP address. At the same time, they struggle the hardest against websites that monitor IP reputation. Using one won’t necessarily get you blocked; but the threshold will be much lower and further modified by factors like ASN (company that owns the IP) and usage history.
We modeled our benchmarks around three questions:
- What’s the success rate with popular targets?
- How fast are the proxies to establish a connection?
- Does an IP database identify them as proxies?
Blazing SEO and Oxylabs aced the tests, others encountered some issues. Our assortment of websites revolved around e-commerce, which is a popular use case for datacenter IPs. Overall, the providers gave us good proxies. Others found targets like Amazon and Craigslist a challenge: for example, Smartproxy had 27% and Bright Data 42% of their IPs blocked on Craigslist.
Otherwise, the proxies had very few connection errors. Indeed.com was an exception, but that’s probably the website’s way of blocking unwanted addresses.
Datacenter proxies are fast, but not equally so. Overall, it took the datacenter proxies an average of 1.75 seconds to reach their destination. This is significantly faster than residential or mobile IPs. IPRoyal’s connection proved to be the fastest, and Bright Data’s slowest. In some cases (see: Craigslist) the latter connected half as fast. Maybe the server hosting the IPs was under more load, or simply slower.
Dedicated datacenter proxies easily reach 100 Mbps speeds. At least four out of the five providers did. Bright Data was several times slower, but its proxies should be fine for streaming video or downloading large files.
Proxy detection status had little impact on the results (or we used a wrong IP database). We ran each proxy through the ipapi database to check if they had been identified as hosting or proxy IPs. However, even when they were, that didn’t seem to have an impact on the results: Blazing SEO did very well despite a half of its IPs appearing as proxies. So, either the databases that our target websites use had different data (which is likely), or the targets decided not to block IPs based on this data point alone.
Like all proxy types, dedicated datacenter proxies are about location coverage, protocols, and authentication options. But they also involve specific features that are less relevant for residential or mobile IPs.
For one, dedicated datacenter proxies come in finite static lists. Clients don’t expect them to rotate, but they do ask for the ability to refresh blocked addresses. Second, these proxies are marketed as unlimited: carrying no limitations on threads, traffic, and domains (aside from blacklisted websites).
For location variety go with Bright Data or Oxylabs. Both providers allow selecting the most countries, several locations per plan, and even particular cities. Bright Data’s system is the most comfortable, as it supports self-service; Oxylabs requires talking with an account manager. Blazing SEO allows choosing cities in the dashboard if the location supports it.
Getting SOCKS5 proxies won’t be a problem. Four out of the five participants offer the protocol as an option. It works great for datacenter addresses, as they have high throughput and no traffic limits. Smartproxy promises to introduce the feature soon™.
Premium providers are stingy about replacing blocked IPs. Bright Data charges for each replacement, while Oxylabs has a threshold how many IPs clients can burn. IPRoyal and Smartproxy, both of which target smaller clients, can replace the whole proxy list monthly, no questions asked. Blazing SEO is highly flexible in that you can change individual IPs you don’t like.
Most services are truly unlimited – Bright Data’s can be if you shell out the cash. Four out of the five providers deliver impose no limits on traffic, threads, or domains clients can access. Bright Data does, but it also has an option to remove them. As we’ll see, limitations make a part of the provider’s business model.
Datacenter proxy services are usually straightforward about pricing: buy a set number of IPs, pay per unit, use them to your heart’s content. But that’s not always the case. One provider we’ve tried allows customizing nearly everything, from traffic to threads, refreshes, and even connection quality. There’s a lot of room for variations, even if the market chooses not to apply them.
Let’s have a look at what strategies the participants choose and how they position themselves.
All participants have straightforward pricing models – except for Bright Data. The company’s pricing structure is as much customizable as it’s complex. Clients can choose the number of domains they want to access, enter their traffic needs, and enable the ability to target particular cities. Naturally. each feature bloats the price, to the point where a barebones configuration costs 2-3 times less than the full one.
Some providers charge more for non-US proxies. Blazing SEO and Oxylabs are a case in point – their proxies outside of the US are 50% or more expensive, based on the location. The others don’t follow this practice.
Each participants positions itself differently:
- IPRoyal goes cheap and targets smaller clients – while you can buy more than 100 IPs in the dashboard, the price per IP remains the same.
- Smartproxy also targets bargain hunters; but its plans start higher up and continue scaling to thousands of IPs.
- Blazing SEO tries to capture all potential customers with broad plans and mid-level prices that scale rather slowly.
- Bright Data aims for a wide range of clients as well but does so through customizability. It ends up favoring better-paying users that can better rates by committing to spend set amounts.
- Oxylabs makes itself approachable to premium clients only with plans that start from $180.
Overall, IPRoyal and Smartproxy can be treated as affordable providers, Blazing SEO as a mid-level provider, and Oxylabs as a premium one. It’s hard to position Bright Data; but, giving its proxies the same features as the competitors makes it the most expensive option (at least when buying US IPs).
So, where do the participants stand next to one another? Here’s a radar graph with our interpretation:
All in all, IPRoyal and Smartproxy are value-oriented providers. They compensate for limited features – mostly in terms of location targeting – with lower prices. This makes the providers a bargain for clients that are content with the limitations, and a less suitable choice for those that need more. Smartproxy seems to scale better of the two.
Blazing SEO’s graph is highly balanced, and we believe that to be the case. The provider offers a respectable number of locations and features, prices that aren’t the lowest but cover the whole range, and great performance.
Bright Data and Oxylabs steer toward the premium end of the scale. Both have an impressive number of features, and both cost above average. Overall, Bright Data provides more customization options, which come at a price. It’s just a little unnerving that the proxies we received were outperformed even by the entry-level providers. Oxylabs performed very well, as expected from a premium service.
Residential proxies are a second popular option for proxy servers. They’re used when datacenter IPs can’t access targets due to IP reputation checks, or when granular location targeting is needed. With the web becoming more gated and geographically tailored, this covers more use cases with each passing year.
Our residential proxy evaluation is perhaps the most complete out of the four services. First, because it includes all the largest providers (truth be told, our selection procedure was designed around this proxy type); and second, because we’ve had several years to refine the methodology.
You’ll find this section divided into five parts:
- Proxy pool, which evaluates the size and composition of each proxy network,
- Proxy performance, which benchmarks how robust the providers’ infrastructures are,
- Features, where we compare what the participants can bring to the table,
- Price, which evaluates how much the first three parts cost, and finally
- Wrap-up, where we briefly summarize our findings and draw conclusions. You can jump straight to it by clicking here.
Residential IP pools are in constant flux – their availability changes hourly, as proxy peers connect and disconnect their devices from the network. It’s very hard to capture such a pool’s size over time, so the numbers that providers advertise are monthly averages at best. They’re often inflated for marketing purposes and always change from one month to the other.
At the same time, a residential proxy network’s size is the most important criterion in choosing a service. No matter how good the performance is or how many features they can get, clients will always prioritize having enough unique IPs in locations they care about.
Our residential proxy pools tests measure:
- How many unique IPs the providers have globally.
- How many unique IPs they control in seven high-value locations.
- What percentage of those IPs is really residential.
Some advertised proxy networks are 20 and more times larger compared to smaller competitors. With 100 million monthly IPs, Oxylabs controls the largest residential proxy pool in theory. Compared to GeoSurf’s 3.5 million IPs, the difference is gigantic.
Providers with the largest IP pools can back their claims; the others – not always. In our tests, Bright Data, Oxylabs, and Smartproxy returned 2-20 times more unique IPs than the competition. NetNut also performed well. However, providers like PacketStream and RSocks could barely beat GeoSurf despite advertising several times more IPs.
Shifter and especially Blazing SEO were the furthest from their claims. We can understand Blazing SEO, as it’s a barely out-of-beta service. Shifter, however, underperformed.
A large proxy pool doesn’t guarantee IPs in high-value locations. Not always, at least. Providers with the most IPs overall also performed well with specific country pools. PacketStream, RSocks, and SOAX, on the other hand, had a disproportionate number of Russian addresses and few US or European proxies. GeoSurf overperformed in comparison.
Furthermore, PacketStream and IPRoyal had nearly identical results. We don’t want to jump to conclusions, but the evidence strongly suggests reselling.
Unfortunately, the number of ports we received from Shifter simply wasn’t enough to complete these tests on time.
Most providers really offer residential addresses – but some tend to cheat. Proxies in the US were the worst offenders. Over 30% of NetNut’s IPs were classified as datacenter; the provider likely added its ISP proxies into the pool. GeoSurf surprised with over 15% datacenter (and 5% educational) IPs, which is uncharacteristic for this provider and something we didn’t expect. Shifter also had a fair share of datacenter proxies mixed in – be it the US or UK.
The residential proxy performance benchmarks answer four questions:
- How reliably the proxy infrastructure can connect to a target (success rate).
- How long it takes to establish a connection (response time).
- How well the infrastructure handles load (stress test).
- How these results translate to popular targets like Google and Amazon.
The first two were made together with the proxy pool tests; their scale makes the benchmarks highly dependable and resistant to outliers. However, the target tests shouldn’t be taken for granted: we made much fewer connection requests, and the results depend on the web scraping setup.
10 Out of 11 providers are able to achieve over 90% success rate. This shows that nowadays proxy companies have robust infrastructure. Only RSocks dropped the ball, which is strange considering that we recently tested the same pool and got nearly a 99% success rate. But the scope was smaller (250k vs 1 million connection requests), and maybe things have changed since then.
This year, it’s hard to single out a leader, as Blazing SEO, Bright Data, Oxylabs, Smartproxy, and SOAX all linger around the 99% mark.
Response time depends on IP location and differs up to 3 times between providers. The average response time across the board was around 1.9 seconds. Oxylabs and Bright Data are the obvious speed kings: the former was up to three times faster compared to IPRoyal and PacketStream. Smartproxy and NetNut are close behind. The others gravitate around 2-2.5 seconds.
The response time for individual countries suggests where the providers have their load balancing servers. For example, NetNut was very speedy in the US but not so much in Russia. RSocks and SOAX – the other way around. Bright Data and Oxylabs were amazingly fast in Europe, especially Oxylabs with a 0.5 s response time (!) in the UK. However, it’s hard to explain PacketStream’s sluggishness in the US, as it’s an America-based provider.
NetNut’s and SOAX’s results have improved over year, PacketStream’s declined. Compared to the year before, SOAX has made the biggest improvement in response time and NetNut in success rate. Team SOAX told us they were running constant routing experiments, but we didn’t expect the change to be this notable (3.69 -> 2.34 seconds). For some reason, PacketStream’s success rate has markedly declined – maybe because of high network load.
Most providers can handle running hundreds of requests per second. This year, our stress test didn’t tell much: it failed to make a dent in all but one proxy network. PacketStream’s success rate suffered, but that might have coincided with outside causes. That said, the test did have an impact on response time. Most providers saw it increase, but GeoSurf and SOAX performed even faster than they had otherwise.
Residential IPs rarely get blocked – unless it’s Amazon or Google. The majority of requests failed due to connection errors, primarily timeouts. Our 30 second timeout threshold is pretty strict, but providers like Oxylabs and SOAX didn’t seem to care. However, it did have a noticeable impact on PacketStream, RSocks, and especially NetNut.
The two hardest targets were Google and Amazon. They caused the largest number of blocks in the form of fake 200 codes (Amazon) and 429 CAPTCHAs (Google). For some reason, NetNut’s results were catastrophic with Google, where only 1% of the requests reached the target. We even double checked if targeting Google was allowed. It was. Craigslist caused issues for IPRoyal, and providers faced some blocks by Instagram; otherwise, all did well.
Smartproxy performed the best with popular websites. The provider had shown strong results already in our general benchmarks. But for some reason, it did particularly well with target tests: the success rate remained high and the response time beat all other competitors.
Bright Data gave us a proxy-based API instead, which didn’t always work out. The provider insisted that we use Web Unlocker which adds extra web scraping features onto the proxy network. It was obviously optimized for websites like Amazon but slowed to a crawl against Walmart and Craigslist. This reduced the overall response time by a lot. On the other hand, Bright Data was the only provider with zero blocks. So maybe its strengths would have shown better on a larger scale.
A provider’s features determine how well the proxy network can adapt to particular needs of the user. If an IP pool includes addresses from Italy, but provides no method to target them, this will make it impossible to track keywords reliably or test an application in that location. If the proxy IPs have inconvenient rotation options, it will require extra development effort to customize them as needed. You get the drift.
By features, we primarily mean the following functionality:
- Location coverage and the ability to filter them.
- IP rotation options that either enable the famous unlimited scaling of residential IPs or allow creating sticky sessions.
- Supported protocols, namely HTTP(S) and SOCKS5.
- Authentication options like username and password, as well as creating sub-users.
150+ countries and city targeting are becoming a standard, ASN filtering remains a luxury. Most providers offers country-level out of the box. Only Shifter has this functionality as an upsell (in 2022!). City targeting has become a commodity as well. ASN filtering, however, is a different matter. Out of the few providers that offer it, only SOAX does so for free. Bright Data, on the other hand, charges double (!) the regular price.
The question whether to allow filtering ASNs is as much political as it’s technical. Small proxy pools simply don’t have enough IPs to ensure a good user experience. Providers realize that and are cautious. It could also be the case of protecting the proxy network.
Everyone supports sticky sessions, some less flexible than others. There are three approaches: as long as available (PacketStream, NetNut), fixed time intervals (Oxylabs, Smartproxy), or highly customizable (SOAX, IPRoyal). Newer providers tend to prefer the third approach. We’ve also seen proxy sellers that force rotation after a specific amount of requests, but none of the participants practice this (except maybe for Bright Data with its Proxy Manager).
On the other hand, rotation on every connection request is almost universal. Shifter’s method is the most limiting, as it charges for ports and only rotates IPs every five minutes. It caused us to omit some of the tests because even with a $10,000 plan, we couldn’t complete them on time.
Nearly half of the providers still don’t have SOCKS5. More particularly, five out of 11 providers allow connecting via HTTP(S) only. It’s interesting that most of them are older companies; we could assume they either haven’t found the need to implement SOCKS5 or decided against it after calculating the benefits and risks. Newer providers think otherwise: they both support SOCKS5 and make it easily accessible (unlike Bright Data, which requires additional software to enable the functionality).
Both authentication methods are common, sub-users come with reseller plans. Only PacketStream fails to support IP whitelisting. Shifter and RSocks have it the other way around: that’s the only authentication method they offer. Knowing that it’s not ideal for users with dynamic IPs, both have implemented workarounds. Shifter gives 100 whitelist slots hoping that they’ll be enough, while RSocks has built a piece of software that periodically re-authenticates the address.
Providers aren’t keen on allowing sub-users for their regular clients. The ones that do mostly serve premium customers that are likely to need proxies for multiple projects. Having separate users helps to keep their traffic use and expenses clear.
Resellers are welcome. Eight out of 11 participants support reselling or have white-label programs. Three don’t, but their reasons differ. Whereas Bright Data has an explicit no-reselling policy, RSocks probably can’t resell because it’s already a reseller and the source forbids it. And GeoSurf, to our knowledge, simply doesn’t allow it.
Residential proxies stand in the middle of the pricing ladder: they’re cheaper than mobile or ISP addresses, but more expensive than datacenter IPs.
It’s impossible to price residential proxies by IP address, so providers offer data-capped monthly plans instead. This isn’t a hard requirement: you can find plans without traffic limits, as well as services that impose no time commitments. They’re simply rarer.
The price of residential proxies depends on how providers position themselves. Companies that target smaller clients will have low entry thresholds and cheaper prices per gigabyte. Premium providers cost more and start higher up, but they also scale well (sometimes even outscaling the cheap providers at terabytes of data). That’s why the market can support so many sneaker proxy and general-purpose resellers.
Let’s have a look at what strategies the providers choose and how they position themselves in the market.
Providers prefer a traffic-based subscription model. It keeps the IP pool cleaner and, from our experience, ensures better performance. Even RSocks, which is famous for unlimited-traffic residential proxy lists, gave us its metered service in an effort to compete in quality. Shifter is the only participant that chooses to charge for ports, while SOAX combines the two approaches.
Pay as you go is still rare. Instead, providers stick to the subscription model. Most offer to buy extra traffic at a fixed rate if a plan proves too limiting but not enough for the next tier. Yearly contracts cut the price by around 15%. Pay as you go is mostly used by the affordable providers, PacketStream and IPRoyal. Bright Data cleverly combines both models: you can pay as you go at a premium rate until you reach cheaper monthly commitment thresholds.
Better-paying clients get more features. For example, Bright Data charges twice (!) for city and ASN targeting, while Smartproxy sells more sub-users. Many of the extra features are unlocked by getting larger plans. NetNut is the most obvious example, as it not only enables city targeting, assigns an account manager, but also provides access to a web scraping API. A curious choice not to sell it as a separate product.
Providers fall into three pricing tiers. They are:
- Cheap: PacketStream, IPRoyal, RSocks.
- Mid-level: Smartproxy, SOAX, Blazing SEO.
- Premium: everyone else.
The difference between these pricing tiers can be staggering. At 50 GBs of traffic, Bright Data costs 12.5 times more than PacketStream.
However, the gap shrinks as the usage scales up. For example, NetNut highly incentivizes volume use, and it starts picking up at 1 TB of data and further. Smartproxy and SOAX both strike a good balance, though SOAX’s plans are best suited for small business use.
It’s interesting – and pretty bizarre – that most of the cheapest providers offer terabyte-scale plans and even encourage reselling. It’s even more bizarre that one of them likely does resell the other. Their limited pool sizes and low prices don’t bode well for IP quality.
There’s no one approach for letting customers test residential IPs. Trials may seem like the most popular option, but they’re preferred by premium services and handed out to businesses. They can also come on condition: for RSocks, you have to leave a good review on TrustPilot; for SOAX, the trial is paid ($1.99). PacketStream and IPRoyal officially don’t offer trials or refunds, though the former does hand out credit on forums like BlackHatWorld.
Here is our visual interpretation of the results. For your convenience, we’ve also divided the providers into cheap, mid-level, and premium based on their pricing:
The graphs are telling: most providers, even the cheapest ones, perform relatively well. While premium companies like Oxylabs and Bright Data still lead the pack, there are no longer huge differences in how well the infrastructure works. Features are mostly not an issue, as well: SOAX and IPRoyal bring much of the same functionality at a lower price point.
The real differences reveal themselves in the proxy pools. As a rule, premium providers had both larger and better-balanced IP networks. The disparity in numbers between Oxylabs and PacketStream reached 2.5 times in all locations, and more than 10 times counting US-only addresses. This will have a noticeable, whether in IP quality or experience targeting less popular locations.
PacketStream has gained strong competitors in the affordable segment. For now, IPRoyal basically sells the same proxy pool; but the provider offers more features and has laid the groundwork to grow its own proxy network. RSocks’ service is surprisingly competitive for the price.
Overall, Bright Data and Oxylabs still have the best residential proxy services, and Smartproxy remains the best value choice. Despite some imperfections, SOAX and NetNut continue being great alternatives in most areas. SOAX offers excellent functionality free of charge but is held back by its proxy pool, while NetNut suffers from minor imperfections in several areas.
As for the others, GeoSurf remains a viable premium choice – it had a pretty balanced proxy pools but didn’t stand out much. Blazing SEO hasn’t left the beta; and despite its respectable performance, the provider’s proxy pool at this time is much too small for any larger tasks.
Few providers offer peer-to-peer mobile proxy networks. They’re hard to source, have data use restrictions, and cost a lot of money. Providers, mostly smaller, turn to dongle-based mobile IPs to counteract these issues. Still, mobile proxy pools have their uses, especially when there’s a need for location variety or larger-scale automation.
Three out of the four providers in the research base their mobile proxies on peer-to-peer networks. Blazing SEO currently uses a different approach: it mixes dedicated devices with IPs borrowed from other people. As we’ll see, this method works but has some implications for a web scraping-oriented service.
The section mostly mirrors the structure of residential proxies. You’ll find it divided into five parts:
- Proxy pool, which evaluates the size and composition of each mobile proxy network,
- Proxy performance, which benchmarks the success rate and response time metrics,
- Features, where we compare what the participants offer,
- Price, which address cost considerations, and finally
- Wrap-up, where we connect the dots from the first four parts. You can jump to it by clicking here.
Peer-to-peer mobile proxies follow all the same conventions as residential IPs: large, fluctuating proxy pools with IPs that come and go. The main difference here is that the proxy peers are connected to mobile carriers and not fixed line networks.
Considering this, we made all the same benchmarks, only on a slightly smaller scale. They measure:
- How many unique IPs the providers have globally.
- How many IPs they control in seven high-value locations.
- What percentage of those IPs is really mobile.
Oxylabs advertises the largest mobile proxy pool. In theory, it’s nearly three times larger than the closest competitor Bright Data, and more than five times the size of SOAX.
Blazing SEO has no number beside it. Why? Dongle-based mobile proxies follow different rules than peer-to-peer networks. They have few IPs available at the same time – as many as there are devices. But they can force mobile carriers to assign new IPs by restarting the devices, which results in thousands of unique IP addresses over time.
Advertised IP numbers don’t always reflect real proxy count. Oxylabs did have the most unique IPs throughout all locations, but the difference wasn’t as big as the advertised numbers would suggest. What really surprised us, though, was that SOAX actually returned more IPs than Bright Data. It stands to reason that we shouldn’t take the official numbers for granted.
Oxylabs dominates high-value locations, Bright Data underperforms. Benchmarks with individual locations brought out the differences. Oxylabs clearly dwarfed the other providers, especially in the US. Blazing SEO had a fair number of American proxies, too – frankly, more than in its residential pool. But remember: very few of the IPs are available at the same time.
However, Bright Data once again disappoints. SOAX clearly had more proxies to work with, even without its large Russian presence. In the US, the difference was more than double. Either we caught Bright Data during a bad spell, or one of these providers should update their IP numbers.
Providers don’t lie about selling mobile proxies. There were no alarms and no surprises. All four providers seemed to be truthful about their IPs being mobile.
The mobile proxy pools had to go through similar benchmarks as the residential IP networks. They answer two questions:
- How reliably the proxy infrastructure can connect to a target (success rate).
- How long it takes to establish a connection (response time).
Once again, the benchmarks were based on hundreds of thousands of requests, so their scale allows making pretty confident assumptions about the services. Only our target tests that accessed popular websites were made on a smaller scale.
Three out of four participants had nearly perfect success rates. Compared to our experience with mobile IPs years before, providers have perfected their mobile proxy infrastructure to work nearly without fail. That might be an exaggeration, and I’m sure that customers have their stories. But in general, the services have improved greatly.
Blazing SEO was the only participant out of the 4 that had some hiccups. But it’s also a very new product for the provider, so these issues will get sorted with time.
Some providers are two – and even four – times faster than others. Mobile proxies connect to websites around half as fast as residential IPs. But whether you’ll find them adequate or slow highly depends on the provider. For example, Oxylabs’ response time is faster than seven out of the 11 residential (!) proxy networks. Bright Data and SOAX were slower but still decently fast. However, we can’t say the same about Blazing SEO: Oxylabs beat it by four, Bright Data three, SOAX by two times.
All providers displayed the best results in Europe. Bright Data favored France, Oxylabs UK and Germany, while SOAX had the best speed in Russia. Oxylabs was the only provider with a response time of fewer than two seconds in the US.
Connection errors – not blocks – cause mobile proxies to fail with popular websites. The mobile proxies didn’t fail often. Whenever they did, the cause was predominantly connection errors, not blocks. The only website that blocked some requests was Amazon. But even then, the numbers were in single digits and could’ve been caused by our web scraping setup.
What the tests did, though, was intensify the issues that the proxy networks already had. For example, Blazing SEO’s response time increased even more, and timeouts caused one third of the requests to fail. The other providers had few flaws, so they experienced little change.
What makes or breaks a mobile proxy service is its ability to offer particular locations. IP rotation is a second important characteristic that determines whether the proxy network is suitable for specific use cases. Protocols and authentication options follow.
150+ countries and ASN filtering are the standard. Aside from Blazing SEO, providers offer over 150 locations to choose from. The pool sizes aren’t that big, so we should be realistic about some locations (like Australia), but there should be at least a few addresses in most of them.
Note that fewer providers allow targeting cities than they do ASNs. The combination of city plus ASN targeting is the golden standard for tasks like account management – websites tolerate IP rotation as long as new addresses have similar parameters. So, it’s strange that Oxylabs doesn’t allow targeting cities.
Bright Data and Oxylabs have the most versatile rotation settings. Three out of the four providers support rotation with every connection request. SOAX doesn’t, which limits its web scraping potential. It doesn’t help that the pricing plans (we’ll get to them later) restrict the number of IPs you can access at once to 300-600.
On the other end, Blazing SEO is the only provider that fails to support sticky sessions. At this point, it’s a purely web scraping-targeted service. SOAX and Bright Data (with Proxy Manager) have the most flexible session options, while Oxylabs is somewhere in-between.
Most providers are SOCKS-friendly. Oxylabs is the only company whose mobile proxies can’t use SOCKS5, so it might not be suitable for use cases that absolutely depend on the functionality.
Authentication shouldn’t be an issue, sub-users may be. Bright Data, Oxylabs, and SOAX all support both authentication methods. Blazing SEO has only credentials for now, which should still be enough for most use cases. However, only two providers allow creating sub-users for regular clients. SOAX reserves the functionality for resellers, and Blazing SEO doesn’t offer it at all for its mobile proxies.
Mobile proxies cost the most out of all proxy types, sometimes by a large margin. This is impacted by their sourcing difficulties and the fact that mobile carriers still offer traffic-bound plans. The end users who participate in the proxy networks have to find it worthwhile to spend their data allowance this way. Dongle-based services face another issue – they have to cover unlimited-traffic data plans all by themselves.
Let’s have a look at what strategies the providers choose and how they position themselves in the market.
Traffic-based subscriptions dominate. We’re reviewing mostly peer-to-peer services, so there’s little surprise that they’ve chosen traffic-based models. In fact, we couldn’t name a provider from the broader market that offers unlimited data plans. At this time, only Bright Data offers the ability to pay as you go. However, it costs so much (see below), and the other providers’ plans start so low, that the option hardly seems worth it.
Dongle-based services, on the other hand, nearly always charge by ports (devices you can access). Their primary use case is multiple account management, which requires few IPs at the same time but devours traffic. Blazing SEO is unique in this regard, as it repurposes the tool for volume use. By doing so, it can charge for traffic as well.
Mobile IPs are expensive – but price differences between providers aren’t always large. Maybe because they’re so hard to maintain, mobile proxy services have much lower deviations compared to residential proxies. Sure, Bright Data costs twice as much starting on. But that’s because at 5 GBs it still offers the pay as you go rates. The difference decreases as the monthly plans kick in.
Still, it’s possible to distinguish SOAX and Blazing SEO as cheaper, mid-level providers. Their entry plans are smaller, costs slightly less, and end sooner. Oxylabs is the more affordable premium service, but even its largest plan only reaches 100 GBs of data. It shows that mobile proxies aren’t meant to achieve the same scale as residential or datacenter IPs. They’re the fallback option, used out of necessity.
Here’s our interpretation of the results:
Bright Data, Oxylabs, and SOAX all draw similar shapes: they reach toward the proxy pool, performance, and feature aspects, mostly ignoring the price. The difference is how far they go.
Overall, Bright Data has the most to offer in features, and it also performed well. However, the provider’s proxy network was smaller than we expected, and it costs the most out of all providers. SOAX had comparable features and even a slightly larger proxy network but couldn’t quite keep up in performance. Oxylabs convincingly beat both competitors in all areas.
The only graph that stuck out was that of Blazing SEO. The provider significantly outpriced everyone else, and its unusual approach to forming the proxy network had strengths. But the mobile proxies still lack in features, and their performance at the time of testing was very much lacking. That’s said, we’re curious to see how Blazing SEO will improve its now-beta service in the future.
Proxy-based APIs have been around for a while. But, with the exception of Crawlera (now Zyte’s Smart Proxy Manager), few providers offered them directly. Instead, they’d supply infrastructure (read: proxy IPs) to companies that specialized in such tools.
This all changed several years ago when more proxy providers started creating their own APIs. Currently, not only do they sell these services alongside proxy networks, but APIs sometimes become the only way to access certain websites, such as Google. You can see it with Bright Data and Oxylabs, both of which dictate the market’s direction.
To keep the scale manageable, we focused on Google as the main target and chose the APIs accordingly. We hope to revisit the topic in a more comprehensive manner later.
This part comprises five sections:
- Performance, where we test how accurately and quickly the APIs are able to retrieve results,
- Features, which investigates integration options and other important considerations,
- Parsing, which shows how well the tools are able to process the data they collected,
- Price, which compares how much the services cost, and
- Wrap-up, where we summarize the findings. Jump to it by clicking here.
With proxy-based APIs, providers aim to reduce the complexity of web scraping to sending API calls and receiving data. So, their defining feature is the ability to successfully complete requests 100% of the time – no blocks, no errors.
To ensure it, proxy providers apply various techniques in the background: rotate between IPs, stagger request frequency, and automatically try again if data retrieval fails. This can take a while. Accordingly, the second important characteristic is the time it takes to return results.
Finally, large clients care about the allowed rate of data collection, as they may need to get hundreds of pages per minute.
Most providers successfully retrieved data within the time limit. We terminated each connection after 150 seconds, so any errors that occurred were caused by exceeding the threshold.
Oxylabs, Smartproxy, and Bright Data successfully completed virtually all requests within the timeframe. Despite the few errors, Blazing SEO succeeded over 95% of the time. It’s likely that the company’s reliance on datacenter proxies, which normally struggle with Google, could’ve made an impact.
Bright Data was at least twice faster than the competition. Its Search Engine Collector blazed through the requests, completing them in fewer than four seconds. Smartproxy and Oxylabs were half as fast, and Blazing SEO’s tool returned data nearly five (!) times slower.
Only Scraping Robot imposes concurrency limits. It’s 100 requests per minute. To be fair, Blazing SEO claims that the limit is soft – in other words, it can be increased or maybe even lifted. The other providers allow making significantly more connection requests out of the box, to the point where many clients can consider the services unlimited.
Search engine-oriented tools are often called SERP APIs. They’re primarily used by search engine optimization (SEO) specialists to extract organic position data from search engine page results (SERPs). That said, plenty more use cases exist, ranging from analysis of paid ads to brand protection and e-commerce intelligence via Google Shopping.
Google customizes its results based on location and device parameters. So, the ability to customize them is very important for getting accurate results. Other concerns include integration options and delivery formats. For example, large businesses will prefer asynchronous integration, as it markedly reduces resource use at scale.
Three out of the four providers are well suited for local SEO. They support not only country-level targeting but also particular cities. Moreover, the same providers also allow changing the device signature for mobile-tailored queries.
Participants prefer returning data over open connection. Or rather, they haven’t yet developed the logic to send queries and retrieve results on demand. Only Bright Data with Oxylabs support the functionality through webhooks.
Though Bright Data allows all three integration methods, it leans toward proxy-like integration – this is obvious looking at the docs, where the other two methods are only mentioned at the very end.
Only two proxy-based APIs deliver data in batches, one in CSV. This makes sense considering that Oxylabs and Bright data are also the only providers to support asynchronous data retrieval. Oxylabs accepts up to 1,000 queries in one go; Bright Data fails to specify the limit.
Oxylabs is the only provider that supports CSV output in addition to HTML and JSON. To be fair, the feature is limited to certain aspects of web search.
When it comes to Google, proxy-based APIs not only scrape raw HTML – they can also structure the data they collected. In other words, they have built-in parsers. This is a valuable feature, especially for smaller businesses that may lack the manpower to develop a parser on their own.
We consider the ability to parse web search, more particularly organic and paid results, to be of most value. But businesses have plenty of other uses for Google’s data: they may also benefit from collecting product information, news, or images.
Bright Data, Oxylabs, and Smartproxy parse nearly everything, Blazing SEO – only the essentials. The first three providers can process nearly every search module, be it the local pack, knowledge graph, or image search.
With Blazing SEO’s API, on the other hand, we were only able to consistently get organic data, paid results, related queries, and people also ask. The company does provide tools to extract other types of data, but you’d basically have to construct the parser yourself.
Each provider has its own parsing schema. For example, Oxylabs puts organic results before paid, Bright Data lists the URLs of following search pages, and Blazing SEO’s tool tries to extract as many people also ask queries as possible. We have provided samples for comparison.
Overall, Scraping Robot’s parser seems to need some work compared to the others. It doesn’t list ranking numbers and had issues structuring people also ask questions correctly. That said, the others weren’t always perfect either: our sample shows that Bright Data failed to return paid ads, while Oxylabs and Smartproxy omitted the knowledge graph.
Unlike regular proxy networks that sell access to a tool, proxy-based APIs shift the focus to delivering results. Their business model is almost invariably based on successful requests – data that reaches the customer.
Let’s have a look at how the participants structure the pricing of their services and how much they cost.
Monthly subscriptions are preferred but not required. Bright Data offers an option to pay as you go, and Blazing SEO’s credits never expire after purchase. In reality though, you’d want to commit to Search Engine Collector’s monthly subscription, as it halves the price.
The participants occupy distinct positions on the pricing ladder:
- Blazing SEO positions itself as highly accessible to entry-level clients. The service effectively has one rate, but it remains the cheapest until a thousand dollars, where custom quotes kick in.
- Bright Data tries to cover the whole range too, but its pricing hardly makes sense for businesses that don’t want to invest $500 or more. For those that do, the rates remain premium but scale well.
- Smartproxy tries to capture mid-level customers with plans that start lower and cost less compared to the premium alternatives.
- Oxylabs clearly offers a premium service. Its plans are rigid and few because the provider onboard clients via sales rather than self-service.
Here’s our interpretation of the results:
Like always, we have affordable services that skimp on other aspects, big guns for premium customers, and something in-between.
Blazing SEO’s Scraping Robot is an affordable service. Most of its appeal lies in pricing: low rates and no expiration of credits. The API can – and does – deliver essential data, though slower and skimping on features that some clients may need.
Bright Data and Oxylabs both compete for premium customers: their plans and underlying infrastructure are well suited for it, and they have the features that enterprises need (such as asynchronous delivery). Bright Data’s performance is especially impressive, though we’d need more demanding benchmarks to push these providers to their limits.
Smartproxy actually might be the most attractive if you’re willing to make a trade-off: better rates in exchange for webhook and batch requests. If so, then it offers largely the same experience as the premium alternatives.
It’s easy to measure the experience of a single person. The process becomes much harder when you have to account for all users. One may be content with setting up proxies in a particular way, while another will find the process unbearable. Even things like the speed of customer service are hard to quantify: first, because we’re dealing with people and not machines (most of the time), and second, because we can’t possibly contact the support enough times for any statistical significance.
Nonetheless, user experience is important. And it can include measurable metrics like available payment methods and tools that help to work with the proxy networks. How these features are implemented is another matter, and sometimes a checkmark doesn’t tell the whole story. But you’ll have to trust our experience to make up your mind – there’s simply no way around it.
This part covers five sections:
- Subscription management, which looks into the billing part of the user experience,
- Proxy management, which covers the methods for operating the proxy networks,
- Documentation, where we investigate how well the providers are able to document their services,
- Customer support, which takes a look at the channels and speed of customer service,
- Wrap-up, where we draw conclusions. You can jump to it by clicking here.
Proxy providers run software-as-a-service businesses, which requires continuous interaction with the client. This section investigates how the participants approach the relationship. It covers:
- Whether clients are able to access the proxy networks by themselves,
- If the participants support the wallet functionality that allows storing funds on their platforms, and
- Whether clients are able to extend their balance without buying a separate plan by hand.
Most providers have implemented self-service. GeoSurf requires all clients to go through a sales process, while Oxylabs and Blazing SEO offer the feature in partial capacity. For Oxylabs, it only applies to the entry residential proxy plans; Blazing SEO hasn’t been able to develop it yet for its newly released products.
The wallet functionality has become common. It’s a useful feature both for clients and providers. The former can make fewer transactions, especially when a provider offers top-ups for traffic-based plans. The latter can better retain clients by awarding funds on the platform. For example, Bright Data makes good use of the feature by matching new customers that add a certain amount of credits.
However, fewer participants offer automatic top-ups. This feature mainly applies to residential proxy plans. It lets customers add traffic or funds once they exceed the allowance. Shifter finds no need because it serves unlimited traffic, while NetNut only allows automatically buying a new plan instead of extending an existing one.
Proxy management covers the ways clients interact with proxy networks. We distinguish three aspects here: dashboard, API access, and supporting tools.
The dashboard part deals with how hard (or easy) it is to set up the proxy server and whether the provider offers adequate usage statistics. Note that hard doesn’t always equate to bad – advanced users may even prefer a demanding setup if it ensures customizability. As for the statistics, it mostly applies to constraint-bound services like residential proxies.
API access briefly reviews the tools the providers offer for accessing their services programmatically. We’re interested if the APIs are accessible by regular clients in addition to resellers.
Finally, the third aspect, supporting tools, investigates apps that unlock ways to get more out of the proxy networks.
Bright Data and NetNut’s proxy networks are the hardest to set up. Bright Data’s setup flow is actually very well made, but it requires learning new concepts (Zones), and the provided widget for generating endpoints can overwhelm with toggles. NetNut’s flow, on the other hand, is simply inadequate: you have to hunt for hidden authentication credentials and then use static instructions which fail to give all the necessary information (such as available cities).
Moderately hard flows involved some confusion: for example, SOAX’s approach can’t handle working with both authentication methods well. Or, in the case of Oxylabs’ residential proxies, it leads clients to a long-winded documentation page.
The easiest flows provided simple widgets that prepared the proxy network for use in a few steps. Compare NetNut (left) and GeoSurf (right):
Proxy-based APIs rely on interactive playgrounds. Or, at least, they should. These products are more complex than proxy networks, with multiple parameters to choose from. Having a playground, which allows the user to form code blocks quickly and test them in action, really simplifies things.
Out of the four providers whose APIs we tested, Bright Data offers an excellent playground straight in the dashboard. Blazing SEO and Smartproxy have playgrounds in their documentation, though the former’s code failed to work. This confused us and didn’t inspire confidence. The best Oxylabs could do for now was to include a configuration file that required a separate tool (Postman) to work.
Bright Data and NetNut provide the most detailed usage statistics. This is one area where complexity shines. Both providers include visual graphs that not only support custom time frames but also allow filtering use by various metrics: traffic, requests, errors, and more.
On the other side of the fence, there are providers like IPRoyal and RSocks – their usage stats involve little more than a text field with traffic expenditure. PacketStream does have a visual graph, but there’s no way to adjust its default time frame of two weeks.
Most participants offer APIs, but only six make them accessible to all. The others reserve features like fetching data usage statistics and managing sub-users to resellers. It’s inconvenient because having them available to regular clients would reduce the reliance on dashboards.
Of course, enterprises may get different conditions: for example, GeoSurf enables API reports to clients that buy over 2 TBs of traffic. But we suppose it depends on a case-by-case basis.
Extensions and proxy checkers remain the most popular supporting tools. Most extensions run on Chrome, which is understandable given its dominant position. The proxy checkers are primarily offered by affordable providers like IPRoyal and RSocks. Their services include free or entry-level proxy lists of questionable quality that require filtering IPs.
Only three providers offer proxy managers. Oxylabs’ Proxy Rotator works as an upsell for its datacenter plans, while Blazing SEO’s and Bright Data’s cover more services. Both tools offer powerful logging and web scraping features (like retrying requests); Bright Data’s version is more hands-on and customizable.
Oxylabs is the only provider with its own mobile app for Android devices, while Smartproxy is the single provider (out of the participants) to have an anti-detect browser. It’s available free of charge and can be a great selling point for multiple account managers but requires maintenance to remain useful.
Documentation is an important part of user experience that remains useful throughout a user’s life cycle. Potential clients study FAQs before buying a service, follow instruction manuals to integrate them into software after making a purchase, and then turn to the help desk once they encounter an issue. While hands-on customer service can cover most of these aspects, it requires time and resources that could be delegated elsewhere.
This part assesses the comprehensiveness of the providers’ documentation, also listing the types of help docs they cover.
Bright Data, GeoSurf, Oxylabs, and Smartproxy offer the best documentation. We distinguished these providers last year, and they continue to uphold the industry standard. All four have excellent textual and video material covering configuration instructions, integration tutorials, code samples, and advice on the proper use of their services.
IPRoyal and Blazing SEO have many areas to improve. In the case of Blazing SEO, its information on datacenter proxies is excellent, but the provider hasn’t had the time to document its residential and other barely-out-of-beta services properly. IPRoyal has little more than several integration instructions and anemic FAQs that primarily target search engines.
NetNut and SOAX have bettered their documentation since our last research, to the benefit of their customers. Last year, we found their help docs lacking. However, the providers have since expanded them with more content, integration instructions and, in the case of SOAX, informational videos. This has improved the user experience and brought the providers closer to the frontrunners.
Proxy servers consist of many moving parts, and things inevitably break. This is where customer service comes to the rescue, especially if the issue involves something that the help documents cannot or do not help with. And sometimes, we simply need a human touch.
We consider competent customer service to be very important, especially if your service relies on the constant availability of a proxy network. The agents should ideally be available at all times, via channels that ensure quick communication.
The majority of customer support departments work round the clock. Clients of premium services usually have direct access to an account manager and/or tech team during the daytime; after working hours, the support staff takes over by live chat or email.
NetNut and SOAX are the only providers with gaps in their coverage. Though PacketStream does monitor its service 24/7, it only applies to critical issues; all other messages take longer to receive a response.
Live chat is available with seven participants, provides help much faster than email. On average, messages sent via live chat received a reply within a few minutes, no matter the time of day.
The response time to emails, on the other hand, strongly varied by provider. Bright Data, Oxylabs, and SOAX took less than an hour, while PacketStream and GeoSurf responded in more than 12 hours on average.
Customer support agents mainly speak English, use Google Translate for other tongues. Bright Data’s customer service seems to speak the most languages, but we wonder if the provider didn’t simply list the languages its office employees speak.
One provider gave a curious explanation for why it conducts customer interactions in English only: it’s so that other shifts could take over and understand the issue at hand.
Here’s our interpretation of the results:<