it really depends on what you are trying to do with reporting.  In my opinion, tags are NOT the way to go if you are trying to create reporting for your conversations.

Instead, create Conversation Data attributes (make sure they are list and not text style)

Then, you can set values for these fields in your resolution bot answers, or rules or in your custom bots.

This way, you have a separate column for that information in your exports, and its much easier to get.

When you use List style conversation data fields, you can also use those to filter any reports in intercom.  if you use txt style fields, they are not available for filtering.

Topics can automatically be assigned if the conversation matches your criteria, but this doesn’t always tell you what you need to know. 

for example, we have set up the following conversation data attributes:

Bot, account type, primary reason, secondary reason, etc 

So, during various steps of the custom bot, like after they select which account they want to talk about, the bot will set the “account type” field to the correct value based on their response.  

The primary reason is set after they choose from the first set of reply buttons, and secondary is next.  This allows us to understand what path the user went through in the bot before finally getting to an agent.

note that the one challenge with this right now (intercom please fix this!) is that the data exports and APIs have a hard time getting the conversation data attributes out because of the way they appear in the json.  So, if this is your primary source of data for reporting, consider that.

Here, I personally prefer using tags because you can classify with multiple tags, whereas attributes cannot.

I also use tags for macros and can create a report to understand recurring topics—whether they are tasks my team needs to address, product feedback, etc. I also use them for incidents, which helps me analyze the volume in case one occurs.

There’s one thing why Topics have advantage over all the other - and that is that they work retroactively - you can set up topic and Intercom will go through the whole conversations history it has and will associate conversations to the topic automatically. You can not do that with tags and custom attributes.