It seems like every time I do a demo I get asked "So how do I change the button color or add an alternate image?". The short answer is well, you don't.
Foundry went through a lot of iterations in the early days. We tried style changes and imagery changes as it was a no brainer and a must have to build a CRO tool. So what changed? We started testing with real users and we discovered that copy always matters more and it's less "noisy" than a button change. Style and imagery changes just require a lot more traffic to prove they work which means your valuable ad traffic is not getting the lift it could.
Are you saying visuals are useless?
Absolutely not. In today's world of dopamine driven social media creative is king. Good creative ads are the key to success and their design is the reason they are successful, BUT Foundry doesn't do ads. We optimize what comes after.
The ad is the hook. The user has already signaled their intent. Once you have their interest the job is "easier" to a degree. Now you just need to carry the message home and prove to them your product, service, or whatever it may be is what you promised. This is the disconnect. Copy is king in this world.
What the data says
Think about MECLABS Conversion Sequence Heuristic. Motivation and value are king. Good imagery and design removes the anxiety that you aren't spam or fake, but good copy captures the user's intent and connects the ad's intent to the page. The Fogg Behavior Method: Behavior = Motivation x Ability x Prompt. Motivation is the copy, not the visuals or the button color. Unbounce did a study examining 57 million conversions and found that copy has twice the relative influence on conversion than design changes.
I understand copy works best but I still don't understand why you can't do both?
Foundry has one goal. Increase your conversion rate. We're pretty aggressive about it too. We can be because we are constantly iterating and learning and pruning losers and creating new, smarter variations. We want quick useful iteration that causes improvements as quickly as possible because every time a user clicks your ad you spend money, we should be optimizing that spend in the most impactful way that we can.
This turns most A/B testing frameworks on their heads where statistical significance is king and we make changes slowly and only when we are absolutely sure that one is better than the other.
We prefer to use our Multi-Armed Bandit that already knows what works the best, maximizing conversion rate based on traffic while allowing new challengers to peek through and vie for that winning position. We pretty ruthlessly kill off the low performers so that no traffic is wasted.
If we were to add button color changes and background image changes in as possible variations then the process slows down. The Thompson sampling takes longer to be confident because the changes need more traffic to prove their significance. This means your valuable paid clicks are now optimizing much more slowly and whatever possible conversion lift there is moves at a fraction of the speed it could.
We're always tinkering and we'd love to hear your thoughts. In a world where every product feels like it wants to do every possible thing we think value can be found in simplicity. We want a product that works and solves its problem better than anything else can. We hope you'll give us a try!