Optimal use of Adobe target

[gkalyan]

Is there a best practice on how many tests you can create?

[Gokul_Agiwal]

Hi @gkalyan  - Can you elaborate your question please? in what context? 

There is no limit on test - you can create as many as test you want based on your use cases and choose the right activity type. However there is some guardrail depending on which version of Adobe Target product you're using - Adobe Target Standard / Adobe Target Premium 

https://helpx.adobe.com/legal/product-descriptions/adobe-target.html 

Hope this helps. If any more question please post it.

[gkalyan]

Thank you @Gokul_Agiwal 

The link you for the provided link, this is the product limitation doc which is useful.

But my question is geared towards the best practices, when to say these are too many tests to track or too many variables to get a correct picture.

For example, if want to completely change your entire web page, but have different A/B tests for each component of the webpage, when should we say this is optimal for testing and getting a result. Hope this makes sense.

[kandersen · Accepted solution]

Hi @gkalyan 

In a perfect optimization program you never stop testing your components. If you're referring to a best practice for number of components to tests, then I think several factors plays in but the main one being traffic.

You can use a profile script to ensure you split visitors to ensure they don't fall into several activities on the same page, as this will make it difficult to conclude which of the tests has made an impact. I have an example of such profile script here: https://www.linkedin.com/feed/update/urn:li:activity:7061973384699547648 It is the one named 'mutually-exclusive groups'.

When you start splitting traffic you of course lower the number of visitors in each activity, causing it to take longer to reach a conclusion.

I think it is difficult to give a best practice of a number of tests on the same page, as parameters like traffic and potentially also page loading time plays in.

To stick to your example, then an alternative could be to run an A/B test between the two pages - once you have a winner you can start to drill into the different elements (maybe using an MVT) to better understand which of the elements has impacted the initial A/B test.

I'm aware i'm not providing you with any best practices for number of tests, but i hope this helps anyway.

[gkalyan]

Thank you @kandersen this is helpful.