I am looking to understand some specific differences between the methods for integrating 3rd party data into Adobe Analytics. Specifically - a file based integration (using "transactionID") vs webhooks/ postback based integration (using Adobe VisitorID). I want to understand why one method needs a transaction ID while the other needs only Adobe VisitorID; and what is the impact of this difference in the way data is seen.
I believe you are referencing the data sources types here.
The Transaction ID is very different from the Visitor ID. Here are some aspects about both:
The Transaction ID needs to be enabled for a particular report suite
The Transaction ID and Visitor ID needs to be implemented on the page code as well.
The Transaction ID will combine offline hits containing the same ID with the online hits and give you a reference of a transaction that continued offline or preceded one. The hits containing the ID are clubbed and the visitor information and traffic information cannot be co-related here
The Visitor ID overrides all other visitor identification systems that are available from Analytics. It is used to tie offline visitor data to the online data. If you are using Full processing Data source, then all traffic and conversion data can be tied and each row is read in as a hit and is affected by the processing rule and VISTA rules on the report suite.
Also remember, if using Full processing data source for visitor ID, you will be charged per row of data inserted. This is the only data source type that has costs associated. Normal Visitor ID data source will not have costs, but is not as broad in accepting dimensions as the other.
From what I understand per the link above - we can use "customer integration" - and use visitor ID as the key. I want to understand the exact cons of going this route (as opposed to transactionID route which is clearly the better route for CRM/ call center type of data).
The documentation that you are looking is pretty much what we use for data source. I would point out that this is not very exhaustive for the topic.
And yes, the data sources allows both FTP and API. However, note that the API is more of a method of creating the file on the FTP, which then gets picked up by the servers and processed. As for the recommended data source type is not present. Each data sources has it own reason for being implemented. Make sure you are aware of the differences and have thoroughly tested it in the dev report suite.
A few points to keep in mind which I see mistakes happening are:
The FTP should not have more than 50 mb of total size in files at any time in it. If it crosses, the processing is paused, until the size is reduced.
The FIN file should be empty and have the same name as that of the data sources file.
Data once uploaded through the data sources can not be removed.
Yes, it does. Looking to understand some more around this. I am indeed talking about data source based integration. Can we do data source based integration (either FTP or via API) using visitorID? If yes, can you send me links to that documentation? From what I am reading on the Adobe help documentation, it seems that transaction ID is the recommended key for integrating via data sources.