How are users gathering data to improve their custom bots?

Hi there! Our team has been using Intercom for quite a few years and has spent a good deal of time trying to understand the behaviour of our users to improve our custom bots. 

We currently use the “Path Analysis” tab within each custom bot to understand which options are most selected and pull this data into a spreadsheet. A data download feature is something we have previously requested to make it easier for us to record this information though I have seen no traction.

Other than that we use tags that are automatically applied for specific reactions to messages from the bot (e.g., add a “resolved” tag to a conversation when a user selects that their issue was resolved by the bot) and we perform some very manual conversation analysis by reviewing conversations tagged with an unresolved tag (because they went to a live agent).

We found that the conversation topics feature Intercom has does not really get into the nitty gritty details we need nor are we able to distinguish between custom bots or time periods well enough.

Does anyone have any suggestions on how we might go about improving our custom bots? Are there any third-party integrations you currently use?