I explain it this way:
When I want to send Orchestrator all of the message's schema data I will send it as data type properties. The variable name of payload alerts me that I am getting all of the data. Data type properties is an ARRAY of data. Some of it is string based, integer based and some boolean. The message schema determines what the contents of the data is actually sent.
I then will use scripts to extract the data I need to make the orchestrator workflow work properly from the payload variable.
Now on the other slide we see just ONE piece of data being sent - usually you will want to call the variable in orchestrator that will receive it the same name as its schema name [and this part is very important] the data types MUST match.
So the variable machine will hold data of type string because that is what type it is in the schema.
So either method works....the question is if you have 9 pieces of data from the schema that needs to be sent to orchestrator - do you want to define each piece of data as a custom property and create 9 input variables in orchestrator to receive them?
OR
Do you want to send ALL of the schema data and use a script in orchestra to extract all 9 pieces of data there?
I hope that makes sense.