I'm new in usability testing and I need to arrange a usability testing on final version of our website. I'm not sure which method of testing is best in this occasion? Do I need to go to street and show the website to people and see what they think? or do I need to arrange moderate testing. I think I need to arrange moderate testing and choose some task and sit with people and see if they can finish the tasks and ask them what they think about the UI at the end of the session! Please let me know if I'm right or wrong and any suggestion will be appreciated!

Thank you so much.


3 Answers 3


The safest way (perhaps even the most effective way) is to try and launch your website as 'Beta', or perhaps as a soft launch. But before you do that, have your entire team run usability tests on the website, irrespective of their role in the website. You can exclude the Quality Assurance Experts and Testers for this.

The main intention of this is to try and break your website, and fix the issues before it goes live. It is far less embarrassing if it breaks before production.

Have your website made live for your organization's internal server and ask everybody in your organization (or atleast a few many) to try and break it. All websites will break atleast a dozen times, but that is a good thing. The more you fix, the lesser chance of it to break when it goes live and made available to the public.

Next thing you can do is to get a few people, who you know personally, or via your team/ organization, and run usability tests on it. You can increase the radius of users with successful fixes, step by step.

You can also set up a 'Bug Bounty' program, to attract more people to try and break your website. Set up a contest with attractive prizes and keep it lit. You don't have to spend too much on the rewards. Try gifts like Amazon Vouchers and stuff like that.

After you release it live, the ideal case is 'zero' bugs. But this isn't practical. Learn from each issue and come up with better versions.

As Edison said, 'I have not failed. I've just found 10,000 ways that won't work.'


Kudos to the Edison quote.

Big changes?

Test it with a representative sample. Proxy users, even. Try and see what happens.

Small changes?

Do a split-run experiment with a small % of the live traffic. Try it and see what happens.

Tiny changes? Launch it and see what happens.

Just try something. Think. Do. Learn.

There's no "best method" but avoiding making mistakes is definitely the wrong method.


It can be done as an internal process, when the designers, developers and anyone else can sit down and analyze their system and get the results.

Based on these results, the design and/or code can be modified to be in accordance with the changes they all agree on.

The users can then provide information, on whether:

  • the task was successful or not
  • the task could be performed easily
  • Was the experience interesting, engaging or annoying – their feeling towards the software

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.