Using Optimizely to A/B test remarketing pixels?
Is anyone else doing this? Have any tips/best practices to share?
What are you looking to test/monitor with this test? Sales? Revenue?
Would be interested in seeing how you plan to set this up and what format site you have? eCommerce? Lead-Gen?
> What are you looking to test/monitor with this test? Sales? Revenue?
> Would be interested in seeing how you plan to set this up
- create an experiment with one variation (on a 50/50 split)
- add the tracking code in the code editor
- during signup pass the experiment data to the DB in hidden form fields
- write some logic to ascertain name of remarketing network from experiment data
- create a helper function to fire events for only the remarketing network shown to a user
- e.g. after user submits the signup form and redirects to a new page, call the helper function to fire the right pixel, and then execute a script which adds the user to a segment like 'completed_step_1' after page load
- for offline conversions (e.g. when a user converts to paid) use a callback or something to execute a script that fires an event on the first login after subscription state change
> what format site you have? eCommerce? Lead-Gen?
just your standard b2b saas site.
> Then my question is mostly for what networks would you like to run these tests?
@MartijnSch pretty much whomever. but are you asking out of curiosity or is there something more you're trying to get at?
@MartijnSch , ironically, after we did the set up we chose to do a 100% test with a new platform.
are you familiar with the post july 6th "lots of direct traffic from IE 7 on windows 7" debacle ?
we got jammed up by one of the companies mentioned in a related article, but they never reached out to us. their director of engineering said he'd get back to us after looking into it on their side, but the fact that they didn't proactively reach out to us made it super easy to (temporarily?) pause campaigns with them and shift 100% of our remarketing budget to another platform.
> I think there is certainly value to be gained from the insight that a test like this can deliver.
agreed. for better or worse (as mentioned above) we're not actually going to run the test right now. however, i'm looking forward to running some tests in the near future.
what kinds of tests would you run to suss out the better performers?
here're some rough 50/50 split ideas we've considered testing:
- all traffic (already mentioned)
- trial signup completions
- e.g. promote free support accounts, webinars, demos, etc.
- customers who convert to active but haven't triggered some engagement signal
- basically create segments for engagement activities and guide customers down some path(s)