Moving from Tags to Conversation Attributes. Question about Fin handling attributes. | Community
Skip to main content
Answered

Moving from Tags to Conversation Attributes. Question about Fin handling attributes.

  • December 4, 2025
  • 5 replies
  • 352 views
Simon_N

Hi everyone,

We’re currently using Tags as our main way of classifying conversations in Intercom (e.g. product area, issue type, billing, cancellations, etc.). It’s starting to feel a bit messy and hard to report on, so we’re considering moving to a small set of Conversation Attributes instead, and using Tags only for campaigns/one-offs and stuff like that.

I’m looking for advice from teams who’ve actually done this:

  • How did you decide which things should be Attributes vs Tags?

  • Which Attributes do you find the most useful for reporting (e.g. issue type, product area, plan tier, outcome, churn reason)?

  • Did you make any fields mandatory before closing a conversation? If so, how many before it became annoying for agents?

  • How did you handle the transition period? Did you backfill old conversations or just start from a certain date?

We’re also using Fin and I’d love any tips on that side:

  • Are you using Fin to help set Attributes (directly or indirectly)?

  • Do you let Fin apply Tags/intents and then use Workflows to map those into Attributes?

  • Any pitfalls you hit with Fin misclassifying conversations, or attributes getting messy again?

Any concrete examples (screenshots of your attribute setup, field lists, or workflows etc etc.) would be massively appreciated.

Thanks in advance!
Simon

Best answer by Conor

Ah great topic, and sorry this took me so long to reply! My general recommendation (focused on B2B SaaS companies) would be to:

  • use as few conversation attributes as you can get away with, and in general sentiment and product area are great choices
  • I would probably not use more than 4, especially if you’re able to bring user attributes into Intercom
  • I definitely would get Fin to set attributes - and when you combine that with requiring them to close the conversation, it’s usually not too painful for teammates
  • I personally would not bother back-filling for past conversations
  • I typically use workflows and escalation guidance to handle escalations and to pass to the appropriate teams (based on those attributes)
  • Not really a pitfall, but don’t expect it to be a one-shot fix. You’ll need to go through a period of revision of the name and descriptions based on how well they’re being applied by Fin and teammates (this help page has some basic guidance to be aware of)

There are a whole bunch of cases where my advice doesn’t apply though, so your mileage may vary! 

    5 replies

    John Pjontek
    Forum|alt.badge.img
    • John Pjontek
    • New Participant
    • December 4, 2025

    ​@Simon_N Yeah I’m also interested in this. Since transitioning to Fin, our legacy process of tagging conversations manually or with macros has essentially ended. The conversation topics feature is a bit underwhelming so I was thinking this might be a good solution. In my use case I’m most interested in how it could help organize conversations by something like “issue type.”   

    Simon_N
    • Simon_N
    • Author
    • New Participant
    • December 5, 2025

    Exactly, John. It would be fantastic to hear from ​@Nathan Sudds, ​@Conor, ​@Julian Murray on this, if possible!

    Conor
    Super User ✨
    Forum|alt.badge.img+5
    • Conor
    • Super User ✨
    • Answer
    • December 11, 2025

    Ah great topic, and sorry this took me so long to reply! My general recommendation (focused on B2B SaaS companies) would be to:

    • use as few conversation attributes as you can get away with, and in general sentiment and product area are great choices
    • I would probably not use more than 4, especially if you’re able to bring user attributes into Intercom
    • I definitely would get Fin to set attributes - and when you combine that with requiring them to close the conversation, it’s usually not too painful for teammates
    • I personally would not bother back-filling for past conversations
    • I typically use workflows and escalation guidance to handle escalations and to pass to the appropriate teams (based on those attributes)
    • Not really a pitfall, but don’t expect it to be a one-shot fix. You’ll need to go through a period of revision of the name and descriptions based on how well they’re being applied by Fin and teammates (this help page has some basic guidance to be aware of)

    There are a whole bunch of cases where my advice doesn’t apply though, so your mileage may vary! 

    Behind the scenes with Intercom setups from real customers: https://www.youtube.com/@support-stack | Join customer support leaders learning how to future-proof their teams on Intercom: https://www.customersuccess.cx/daily | Get answers to your Fin questions: https://www.customersuccess.cx/fin-questions-answered
      Simon_N
      • Simon_N
      • Author
      • New Participant
      • December 11, 2025

      Hi Conor, thanks a million. That’s useful info. We will definitely start small. I’ll let you know how we get on.

      And if anyone else has any other experiences, please let me know!

        R
        • Ryan Winkler
        • New Participant
        • January 13, 2026
        • use as few conversation attributes as you can get away with, and in general sentiment and product area are great choices


        I’m not sure I quite understand the logic behind this. Why would I not want my data structured?

        If information, sentiment and specific things I’m trying to capture dies in the conversation or ticket, what use is of it to me?

        Do you not join your data across your customer information? 


        While I understand the perspective of wanting to make moving fast as easy as possible, this seems like a good way to go from “I’m just solving tickets” to using the data as insights and leverage to improve as a company.

        Terms & ConditionsAccessibility statement