How to choose the right plan

How to choose between pay-per-test and subscription:

If there’s one thing we’ve learnt from executing over a 1 million load tests and working with over 140,000 people, it’s that you always need to run more than one test.

Therefore, we suggest using our pay-per-test option only in the following situations:

  • As an add-on to your subscription. Simply put, if you need to run a few tests which exceed the limits of your subscription and you do not want to upgrade your plan.
  • If you're new to our service and simply want to give it a trial run before making any commitments.
  • Finally, we realize that some of our customers can not purchase subscriptions due to company policies, or because they are consultants and only need to test on a project-basis. If this is your situation, we will happily create a customized offer that suits your needs. Contact us for details.

    How to choose a subscription level:

    Selecting a subscription level that’s right for you can be daunting, particularly if you’re new to performance testing. Here are some guidelines for selecting the perfect plan.

  • Start by calculate how many concurrent users your system can handle today. The formula to use is:

    concurrent_users = (peak hourly_visits * avg. visit_duration in seconds) / 3600.

    To find your peak hourly visits and average visit duration using Google Analytics, please follow these instructions.

    If you don’t have any historical data to work with, work backwards from target sales/service projections and conversion rates. For example, how many unique visitors will you need each month in order to generate X number of conversions or sales? Once you have this number, apply the formula above to determine how many concurrent users you will need.

  • Once you have determined the number of concurrent users you will need to be able to handle while still delivering an acceptable performance and user experience, the next step is to find your breaking point (i.e. your maximum capacity)

    Our general rule of thumb is to start by doubling (2x) the number you came up with above. The objective is to find out when your system breaks and what happens when it breaks. And if it doesn’t break, then you know you can handle double your current load levels without fail.

  • When it comes to the number of Server Monitoring Agents needed, we suggest that you install an agent on at least one server in each layer of your architecture.

    For example, If you have 20 frontend servers and are confident in your current load balancing setup, you don’t need to install an agent on all 20 machines, you only need it on one - that one server will be representative.

    Moreover, we advise you not to install Server Metrics Agents on the network level (e.g. load balancers, firewalls, network appliances), these metrics will not provide value to your results analysis.

  • Still need help picking a plan? Or maybe you're unsure whether or not our technology will satisfy your demands? In either case, contact us!