[Feature] Add metrics around human vs Fin AI usage

[daniel-nagish]

Is there an existing feature request for this?

• I have searched the existing issues

Describe the Feature

The current model filters to only give stats on human interactions when fivetran is also collecting bot interactions, as more companies leverage fin ai to auto-resolve issues before escalating to a human, it would be useful to measure those interactions as well

How would you implement this feature?

making the following updates:
models/intermediate/int_intercom__conversation_part_aggregates.sql
Add differentiation for FinAI involvement in ticket handling, allowing for more granular analytics
by adding counts for total, admin parts, and bot parts
adding additional breakdowns of all the other aggregate metrics to have for both admin (human) and bot (ai)
ie first_bot_response in addition to first_admin_response

models/intermediate/int_intercom__conversation_string_aggregates.sql
add an aggregation for all types (lead, admin, bot) that touched a given conversation

models/intermediate/int_intercom__conversation_part_events.sql
add a cte for bot events following the logic from admin_events, except changing the filter to author_type='bot'

models/intercom__conversation_enhanced.sql
add in ai_resolution_stage from the conversation_history table to allow for better tracking of that
models/intercom__conversation_metrics.sql
similar to the above, but also add in the admin metadata so you can drill down by support rep vs ai

Describe alternatives you've considered

disabling these models from package and running an overwritten version, but then I am at risk of missing schema changes as fivetran updates the connector

Are you interested in contributing this feature?

• Yes.
• Yes, but I will need assistance.
• No.

Anything else?

No response

[fivetran-joemarkiewicz]

@daniel-nagish thanks for opening this feature request. The request seems very relevant and something I could see being valuable to other users leveraging the Intercom data. I see you also went ahead and created a PR to contribute the updates you've already applied. Thank you!

My team and I will review these updates, but the consideration of the Fin AI component may not be as straight forward as your other Feature Request (FR #61) and we will likely need to do some additional testing and research before moving these updates along. I'll consider reviewing your updates in our upcoming sprint and will reach out if I have any questions regarding the approach to address this feature request.

Thanks!