Home > The Newbie Zone > Questions and Answers

Is this enough data to stop testing landers that perform similar? (6)

08-20-2018 07:17 PM #1 steel rain (Member)
Is this enough data to stop testing landers that perform similar?

Hello everyone,

I'm currently testing 5 different landing pages with a offer that has an acceptable conversion rate, to find the best one.

This are my results:



Having spent U$200; Should I continue reaching for statistic significance (>90% probability of being best), or just conclude that they are all similar and pick one to test more offers?

Thanks in advance.

08-20-2018 10:08 PM #2 matuloo (Legendary Moderator)

or just conclude that they are all similar and pick one to test more offers?
Yup this is exactly the result of your test, none of them stood out ... which is a good think in a way, since you have several LPs that you can run simultaneously to reduce burnout. On the other hand, all of them perform quite poor, but that might be a problem of the offer ... so yes, that would me my suggestion now, test more offers with all the LPs.

08-20-2018 10:45 PM #3 vortex (Senior Moderator)

What matuloo said.

When you see a situation like this where there are 2 or more split test candidates going head to head, you can just pick an arbitrary lander and use that to test more offers.

When you find a profitable offer, you can rotate the other landers back in to reduce burnout like matuloo suggested.

Or, you can test more new landers or variations to get the CR up. But if you're quite confident that the current landers already include the most popular landers that you have optimized for speed, then I would focus solely on testing offers until I find a profitable one.

It doesn't mean you won't be able to massively increase CR by testing more landers or lander variations. Again, it depends on how confident you are in your current landers. If you are fairly confident, then chances are further testing won't yield as much improvement as testing more offers will.

Hope that makes sense! And you have my condolences...it doesn't happen a lot that ALL your landers would go head to head like that. That's a real money-suck.



Amy

Sent from my SM-G930W8 using STM Forums mobile app

08-20-2018 11:04 PM #4 steel rain (Member)

Can't thank you guys enough. Your help is really appreciated.

08-21-2018 04:20 AM #5 servandosilva (Member)

If the LPs are fairly similar Amy and Matej have covered it.

However if they’re different enough I’d go and dive in with a second layer being OS or Browser to see if you can see any pattern there and maybe one lander converts better in an OS than another.

4 conversions each though wouldn’t be enough to decide IMO.


Sent from my iPhone

08-21-2018 06:53 PM #6 matuloo (Legendary Moderator)

Quote Originally Posted by servandosilva View Post
If the LPs are fairly similar Amy and Matej have covered it.

However if they’re different enough I’d go and dive in with a second layer being OS or Browser to see if you can see any pattern there and maybe one lander converts better in an OS than another.

4 conversions each though wouldn’t be enough to decide IMO.


Sent from my iPhone
Yup, this is worth a shot, thanks for brnging it up Servando! It's possible that the landers perform differently for certain targets ... I'm afraid it's not enough leads to clearly confirm these patterns, but if it's the case indeed, you might want to isolate the better combinations and test them more deeply.

Home > The Newbie Zone > Questions and Answers