Split testing allows you to experiment with different designs and track conversions to determine which one is the most effective. Learn how to do A/B testing using Split in this episode.
I've just released the new version of Split (0.4.0, https://rubygems.org/gems/split/versions/0.4.0) that contains the new config option that Ryan mentioned that helps to avoid users participating in multiple experiments at the same time.
Ah, this time you're 2 days late! I used it earlier this week and is great tool indeed. Customer happy that they can do A/B testing and stop/restart testing sessions themselves, which is a relief also for me ;).
Similar to Vincent, I'm looking to use something like this to track in-page actions. I'm imagining that the solution will be to fire an AJAX hit back to the site to an action which triggers the conversion.
I've started work on a JSON api for split, in the form of another small, mountable sinatra app, the code is here (unfinished): http://github.com/andrew/split-api
This will come with a javascript library for easily doing split ab_tests in javascript.
Slightly different and built for Rails, I've released the split_tester gem. It integrates easily with Google Analytics so that you can use Analytics as your evaluation dashboard.
A quick FYI for others that are using engine yard instances. You will want to have redis running on a utility instance and if you follow EY's recommended chef recipe and implementation, add this to your split initializer:
# Connect to Redis using the redis_config host and port
if redis_config
$redis = Redis.new(:host => redis_config['host'], :port => redis_config['port'])
end
I've just released the new version of Split (0.4.0, https://rubygems.org/gems/split/versions/0.4.0) that contains the new config option that Ryan mentioned that helps to avoid users participating in multiple experiments at the same time.
Thnaks for your work Andrew
Thanks Andrew -- looks like a very useful gem.
nice
Awesome - thanks Ryan & Andrew!
Ah, this time you're 2 days late! I used it earlier this week and is great tool indeed. Customer happy that they can do A/B testing and stop/restart testing sessions themselves, which is a relief also for me ;).
I think the short variant of form won't work since there should be password confirmation validation in the model.
I'd love to use this in a JS frontend with a rails backend. Is there a way or what would be the best way to implement it?
I've been using this gem for a while and it's awesome!
Similar to Vincent, I'm looking to use something like this to track in-page actions. I'm imagining that the solution will be to fire an AJAX hit back to the site to an action which triggers the conversion.
I've started work on a JSON api for split, in the form of another small, mountable sinatra app, the code is here (unfinished): http://github.com/andrew/split-api
This will come with a javascript library for easily doing split ab_tests in javascript.
Nice work Andrew!
Slightly different and built for Rails, I've released the split_tester gem. It integrates easily with Google Analytics so that you can use Analytics as your evaluation dashboard.
https://github.com/jhubert/rails-split-tester
So happy to see more people A/B testing their apps.
That's cool, I actually built a javascript only library thats very similar: https://github.com/andrew/split.js
Just a small notice...
In episode notes there is
config/split_config.rb
which should be
config/initializers/split_config.rb
+1
Was easy to debug, but still cost me 10 minutes...
A quick FYI for others that are using engine yard instances. You will want to have redis running on a utility instance and if you follow EY's recommended chef recipe and implementation, add this to your split initializer:
# Load the redis.yml configuration file
redis_config = YAML.load_file(Rails.root + 'config/redis.yml')[Rails.env]
if $redis
Split.redis = $redis
end
Split.configure do |config|
config.db_failover = true # handle redis errors gracefully
config.db_failover_on_db_error = proc{|error| Rails.logger.error(error.message) }
# config.allow_multiple_experiments = true
end
Great cast!
One question though? How do you set the identity of a user like in A/Bingo?
I want to A/B test for loggin in users and identify them by userid.
First sign in through GitHub to post a comment.