John:
We are working with and testing the Collector and Survey123 apps. What’s nice about Collector is that you can publish a Feature Service, which contains featureclasses stored in a GDB on a SQL Server, from your ArcGIS Server. Make a map in AGOL with the feature service, connect to the map in Collector and any features created in the field are committed back to your on-prem geodatabase. Nice! The downside is that if you want to capture may attributes with your features, as Survey123 is capable of, then you need to go through the process of building many domains and coded values in domains, which is tedious and cumbersome, in my opinion. We use Collector just for simple point-line-polygon capture with few attributes.
Survey123 is much easier to build, much more configurable and easier to use. The downside, in my opinion, is that if you make a schema change to your survey after it has been initially deployed to your field staff and they have already gathered some data, you can’t publish your new survey without overwriting you AGOL hosted feature service, thus wiping out your survey data. Hence, be careful and be sure to copy your AGOL survey data to a backup location prior to publishing a new survey that has a schema change. James’ script is great for this task.
My thought is that field staff will use both Collector and Survey123 in the field. Collector to navigate and see project vector data assets if necessary, and Survey123 to gather the laundry list of values required for your form or reports.
Your CSV question – If you use Repeats or a Photos item in your survey, the CSV download contains your tables, but not your relationships between tables, as you get when you download to FGDB. You also don’t get your photos in a CSV if your survey allows the capture of photos. I prefer the FGDB format.
Hope this helps,
Joe