Pages Navigation Menu

Advanced SEO Tutorials

Categories Navigation Menu

How Many Proxies Do I Need To Save Money?

How Many  Proxies Do I Need To Save Money?
Let Others KnowShare on FacebookTweet about this on TwitterShare on Google+Pin on Pinterest

The big question, how many proxies do I need to use? Well if you only care about protection against complaints you only need 1.

However when you use only 1 proxy your IP will be flagged very quickly with the Captcha services, and then rather than sending you nice easy captcha to solve they will send you much harder ones or even stop sending them. So how many proxies should you buy if you want to avoid the hard captcha? Well that is something even we did not know, we have been running with over 20,000 proxies for about 18 months now so hard captcha have not really been an issue here!

We ran OCR solve tests with the following:

  • 3,000 (Zont of the fastest Proxies)
  • 50 Private Squid Proxies
  • 10 Private Squid Proxies
  • 1 IP Only
  • Control Test- 3,000 proxies using Imagetypers to solve

We ran 3 account creations in a row with an 8 minute wait for email confirmation and, where worthwhile, with 50 and 10 threads in SENuke. We had to run the 50 and 10 thread tests on different days to give the proxies time to ‘clear’ any spam flags they had against them.

One important note is that this was done using Proxifier to rotate the proxy, so if there were 50 proxies in the test then all 50 were being used. If you don’t rotate proxies then for each account creation you are effectively only using 1 proxy! This also allowed us to seperate captcha solving from site signup, the same set of proxies were used for site signup in every case. This was done so that sites which have banned all our proxies, which there are quite a few, will be constant rather than variable for each test.

The SEnuke captcha settings were:

For the control Imagetyperz was used with 3 attempts.

SEnuke was set to 2 site retries.

…and the % solve rates… not quite what we expected:

SEnuke Proxies and OCR Test Results

Results in blue were extrapolated
Note that unlike last weeks test we have used a control to calculate the actual percentage of sites that worked with OCR. This is why the solve rates show as higher than before.

We can see here that when running SEnuke slowly and only doing 1 single account creation it makes very little difference how many proxies you have. Also if we look at the 10 SEnuke thread results for runs 1,2 and 3 we can see that when it comes to multiple runs at slow speeds it also makes no real difference if you have 50 or 10 proxies as your results will be roughly the same. We would put this down to the fact that as SEnuke is running slowly the proxies are effectively getting unflagged before the project is complete.

Moving over to the 50 thread test we start to see a bigger advantage to running with more proxies. In particular at 50 threads by the time you are on your 3rd account run it makes no difference if you have 10 proxies or no proxies (1IP) as the results will be the same.

We can also see from this that the base rate for solving is about 54%, that is you will no fall below this level unless your IP is totally banned.

Impact on Captcha Solving Costs

So given the above data lets work out the average solve % rates over 3 account creations:

Average OCR Solve Rates

Again we see at 10 threads there is minimal difference between 50 and 10 proxies, however lets look at 50 threads and the costs of solving captcha with OCR & human solver combined against only paying to solve with a human solver:

Cost Savings on OCR

Captcha solve costs based on DBC base rate of $1.39/1000

To explain the above for 50,000 captcha the costs to solve using just DBC would in theory be $69.50 whilst using 50 proxies and OCR+DBC it would be $26.90. We have to say ‘in theory’ here as that is assuming a correct solve each time for DBC, which is what they claim to offer, in reality solves are sometimes wrong and the incorrect captcha reporting system never really works properly. As such real life savings will be higher than the above.

Again what we see through is at the low end there is minimal difference to costs by using more proxies, if you are a casual user then 10 proxies will see you through and the extra cost of buying more will be wasted. The cross over point between using 10 and 50 proxies is going to be up around 250,000 captcha solves a month (remembering that the more you do in any one day the bigger the difference between 10 and 50 proxies will be).

What we can definitely say though is that OCR will save most Nukers money. If we take the cheapest Captcha  Tronix package at $15.97 you will breakeven on OCR vs Paid results at 35000 captcha solves which is roughly 5-6 Full Monty templates a month. If you are running less than that then simple paying for your solves is cheaper. If you are in this for the long term then a one off cost of something like Captcha Breaker will of course save you even more money.

For those at the higher end, running 10…20.. 30+ campaigns a day then OCR will be a massive savings. Certainly if you are up there spending $500+ a month on captcha solves you should be looking at investing in a decent proxy system and OCR.

Q&A

Q: Why did the 50 Squids at 10 threads get the highest solve for 1 account creation, surely it should be the 3000 Zont proxies?
A: This is down to the way our proxies work, our system has a slightly lower success than regular proxies for any 1 single account creation. However where it differs is that our system has the same results no matter how many threads or account creations we run. If you look at the average solve rates over 3 account creations at 50 threads you see:

  • Zont Proxies = 70.1%
  • 50 Squids = 61.3%
  • 10 Squids = 57.2%
  • No proxy = 53.8%

Other proxy systems get lower success the more you use them, our system gives the same results every time.

Q: I can’t afford thousands of proxies, what should I do?
A: What we can see is that when we run slower the results are pretty similar no matter how many proxies you have. So simply run slower, reduce you threads down to 10 and your OCR solve rates will go up.

Q: How else can I increase solve rates?
A: Add more retries… however note that the more retries you do the more likely you will be flagged. Thus the best thing you can do is run an account creation, then wait a few hours and re-run it. You will get better results running “4 OCR attempts  -> Wait 4 hours -> 4 OCR attempts” then just running with “8 OCR attempts”

Q: Anything else?
A: Yes add in the human solver as a backup, preferably as the last option in Captcha Breaker after your recaptcha OCR. When OCR fails the human solver will more than likely get it correct, this does 2 things:

  • Gives you a site that otherwise you would not get using OCR
  • Marks your IP as giving a correct solve, helping to reduce getting hard captcha in the first place

If you don’t run any human solver then you will only get results on 2/3 sites in SEnuke.

Q: Are your numbers 100% accurate and will be the same every time?
A: No of course now, there are too many variables and other factors to add in. The results are symbolic however, and are representative of the results people should expect to see. Costs are hard to calculate exactly because of things like premium captcha solve hours, reduced solve costs for ‘gold’ DBC members, updated lists in SEnuke altering success rates.

Q: I am using shared proxies, will they be OK?
A: Shared proxies = abused proxies. Consider them to be pretty much the same as running with no proxies, unless you are very lucky, when it comes to OCR success calculation.

Q: This is too confusing… make it simple!
A: OK how about this:

  • <30,000 captcha a month = use 10 proxies from Squid and use human captcha solver or if long term user Captcha Breaker too
  • >30,000 but < 250,000 = use 10 – 50 proxies from Squid and use OCR + preferably human captcha solver
  • > 250,000 = use a lot of proxies and use multiple OCR + preferably human captcha solver too

Discounts

Remember don’t pay full price….

Pobman

 

Related Post

Let Others KnowShare on FacebookTweet about this on TwitterShare on Google+Pin on Pinterest

2 Comments

  1. “”However when you use only 1 proxy your IP will be flagged very quickly with the Captcha services, and then rather than sending you nice easy captcha to solve they will send you much harder ones””

    Well I know this is my hobby but I never new that ! Does it real happen to such low users like us I can understand some Russia out fit doing millions a day but at our level of use ?

    • 100% you will get flagged, just look at the results from using only 1 proxy … thats because within a few minutes your IP is in the list. Now that will clear in an hour or so… but we assume that they are also tracking long term. We have never, and actually cant, really prove that but one thing we can note is that if we run with a clean home IP it takes longer to get flagged than if we run with a previously used proxy.

      The Russian spammers will not be the ones being impacted by this, as they will be using a similar system to what we use (only most likely an illegal version).

      If you have Captcha Breaker just watch it in action… to start with you should get nice clear recpatcha and then after a few minutes it will move to the harder blob captcha. Recaptcha is not the only one to do this, a few of the other services do this. Eventually you will not even get a captcha image instead you get a message telling you Google thinks your system is sending automated requests. That message is not seen in SENuke very often though, you generally need to abuse the system a lot to see that.

Leave a Comment

Web Hosting

Pin It on Pinterest

Share This

Share this post with your friends!