Recently active topics

I’ve noticed FIN closing some conversations because the user didn’t explicitly confirm escalation, even though the conditions for escalation were met. The rules I set were: Escalate if the user shows negative or neutral sentiment. Escalate when the user provides all required information. Escalate when the user requests escalation. The issue is that when a user provides all the information and FIN sends the escalation message “An agent will be with you soon,” it later follows with “Is there anything else I can help you with?”If the user responds with something like “No,” FIN ignores the first escalation message and treats the user’s “No” as a signal to close the chat. Instead of keeping the ticket open for an agent, it ends the conversation.

Today we noticed an issue with Fin failing to respond to tickets. There have been no changes to the workflow but for whatever reason it just hasn’t responded to queries. We noticed the increase in FRT and drop in involvement but looking back we are almost certain this has happened before. It appears to be intermittent and no real obvious answer but curious if anyone else has experienced this and how you resolved. 

On the intercom landing page, we have some pinned articles. If you open the pinned article, and try to tap the "X" button, it will not close the article. However, swiping down will still allow you to close that article. This behavior only occurs on iOS - it does not occur on Android.Some notes:If you go from the landing page and select the "Help" to go to the collection of articles (rather than the pinned article on the landing page), then those articles CAN be closed via the "X" button. This behavior occurred on the latest react native SDK version, 19.1.2 We checked and found we had to regress to 19.0.3 to resolve the problem - however we cannot use this older version, otherwise we will lose the iOS 26 changes.

Hello. I encountered two issues during implementation in the web app:1. After updating the language via language_override, the intercom language is updated, with the exception of the block for composing a new message (the language changes within this block).To change the language of this block, you need to completely reinitialize the connection via shutdown and boot. This works the same in both demo and user mode.   Also, if we were initialized in our application as a demo user and did an update upon login with user_id, user_hash, and other user data, then the user's message history is not loaded, although his name is displayed and if we write an intercom message, it comes from this user. This can also only be solved by a complete reinitialization via shutdown and boot.

Hi Community, I’m successfully sending WhatsApp template through Series, but no conversation is being created in Outbound. Can you assist if I’m doing anything wrong?

Hi. Creating a new topic as the existing ones about this are quite old.I’m new to Intercom (used HubSpot before). My customer support team used to receive all messages on MS Teams. I see that Intercom doesn’t have an out-of-the box integration with Teams. Is there any 3rd party solution for this or has anyone achieved this in a relatively simple way?Thanks!

 Conor Pendergrast (CustomerSuccess.cx)Hi, I'm Conor. I help B2B SaaS teams turn Intercom into a trustworthy, AI-first support system that reduces workload, lifts accuracy, and keeps customers confident. Through CustomerSuccess.cx, I work with leaders who want Fin to be dependable — not something they have to babysit. I’ve been using Intercom since 2013 and have run support, success, and ops teams, so I focus on systems that work in production, not hype or guesswork.What I’ve done so far Director of Information & Communication at a social care company, founder of a social care tech startup, Senior Leader at Expensify, mediocre triathlete. 10+ years working with Intercom, focused on Fin, workflows, and AI-first support systems. Led Fin optimisation projects for B2B SaaS teams big and small, with clear lifts in resolution rate CX score and teammate reply metrics Known for fixing Fin’s failure points, designing safe Tasks, and building data connectors that improve accuracy. Ho

 Most teams switch on Fin and hope for the best. The best teams build a ritual.In this episode of Support Stack, Conor Pendergrast (CustomerSuccess.cx) and Jen Weaver (Supportman) show you how just 15 minutes a day reviewing Fin’s conversations can transform your AI performance and customer outcomes.What you’ll learn in this episode:🔍 Where to find the conversations Fin couldn’t resolve🛠 How to turn those conversations into better answers with snippets and articles⚖️ Why snippet sprawl is a hidden risk — and how to avoid it📈 The habits that help Fin (and your team) get smarter every dayIf you’re using Intercom and want to get serious about AI-powered support, this conversation will give you practical steps you can use immediately.Subscribe for more behind-the-scenes looks at how real teams are making Intercom work in the wild. 

 Guessing where to improve Fin wastes time. Topics Explorer shows you the exact conversations where Fin struggles so you can fix what matters first. In this Support Stack episode, Jen Weaver (Tettra) and Conor Pendergrast (CustomerSuccess.cx) walk through:Selecting the right scope and metric to surface real opportunities Generating a Testing Group from past conversations by topic Rating answers with keyboard shortcuts and product-expert notes Deciding whether the fix is content (articles/snippets), data (attributes/connectors), or actions (Fin Tasks) Preventing wrong-product answers by pushing the right user attributes into Intercom and guiding Fin accordinglyYou’ll see a real example where a new product area had a low resolution rate due to thin documentation and missing signals. We show how to diagnose it and plan the quickest path to better answers. Perfect for support leaders and ops folks who want a pragmatic, measurable way to lift Fin’s resolution rate without boiling the ocean.

Tired of getting customer messages that just say “It’s broken”? In this episode of Support Stack, Jen Weaver from Supportman shows Conor Pendergrast (CustomerSuccess.cx) exactly how they trained Fin to clarify error details before the ticket ever hits an agent.We go behind the scenes in Intercom’s Fin guidance settings so you can see the wording, the workflow, and the real impact this setup has on reducing back-and-forth.Practical, real-world Intercom examples — no theory, no fluff. 

 What changed when Pupil Progress pushed Intercom’s Fin to the front of their support flow and rewrote their help centre for humans and AI? In this episode of Support Stack,  Conor Pendergrast (CustomerSuccess.cx) interviews Tobi Davis (Head of Customer Experience, Pupil Progress) about how Tobi’s team lifted Fin’s resolution rate from ~55% to ~75% in about two months, and why it now holds steady at 75–80% weekly. You’ll learn: Why going AI-first with a clear “talk to a human” option lets users self-select the path they want How to rewrite your help centre for two audiences: people and Fin The role of audiences and attributes so answers change by role, subject or context When to use custom answers, snippets and narrow articles for precise guidance Why they focused on content and data before actions or procedures How tools like CX Score and Topic Explorer guide the next set of improvements Who this helps: Small support teams trying to scale quality without growing headcount

We would need to add Content Security Policy to our application. I saw this article on Intercom Help. There it is stated to use 'unsafe-inline' but I am wondering is there a way not to use it, since we would like avoid that in our policy?

Hi, im trying to send a message when a client does not answer to a bot. I´m sending a few options in buttons when the conversation starts, and I would like for the bot to identify if the customer hasn´t responded in 30 minutes and send them a “Are you still there?” message, then wait other 30 minutes and close the conversation if there still hasn´t been answered. Is something like this possible? Fin tells me you can´t send automatic messages if the client hasn´t responded to a conversation with the bot.

We’re hoping to gain insight into how other customers are forecasting handle time, particularly when using Intercom data. We've noticed discrepancies in the handle time metrics reported by Intercom that don’t align with the structure of our operations.Typically, I export the raw data and filter out outliers manually. However, we continue to see a significant number of conversations with unusually long handle times that lack a clear explanation. This inconsistency makes it difficult to trust the accuracy of the reports.We’re looking for a better approach—either a clearer understanding of how Intercom calculates handle time behind the scenes, or alternative methods for forecasting handle time more reliably using Intercom data.Thank you in advance for any guidance or shared experiences!

I think it’s absolutely ridiculous that when I receive an email where the sender removes people from the thread and I reply, Intercom adds those people back in, with no notification to me that it is doing this. For example, I am in an email exchange with Julie and Susan. Julie responds to my email but removes Susan from the thread. When I respond, Intercom automatically adds Susan back without 1) informing me that she was removed or 2) giving me the option to keep her off the next email. The only time I know to remove her is when the email from Julie says “Susan moved to BCC” or “minus Susan” or something and then I have to manually remove Susan from the response to not add her back in. This is an issue for a few reasons.1) Customers have gotten frustrated with me that I “keep adding people back in” to the conversation once they’ve removed. I don’t realize I’m doing this because if there are 8 people on a thread and they remove someone without telling me explicitly, I don’t know that t

Discover  how to Use Fin Attributes for Escalations in Workflows & Tasks from Intercom Community Top Expert ​@Nathan Sudds 💡  

Many people rely on disability benefits to maintain stability, and understanding the process can be difficult. In the middle of these challenges, Supplemental Security Income (SSI) offers essential financial support for individuals with limited income, helping them manage basic living needs.

Hi,I am trying to use Custom Objects to show the orders of a users that message us, either through email or whatsapp. I created the custom object Order with the relevant attributes. I then created an external reference in People and Conversations, referencing Order via a one to many relationship. Then I created a webhook that triggers when a conversation is created or a reply is given. The webhook calls a service that posts to /custom_object_instances/OrderWe post an external_id and custom_attributes according to https://developers.intercom.com/docs/references/rest-api/api.intercom.io/custom-object-instances/createcustomobjectinstances -d '{ "external_id": "123", "external_created_at": 1392036272, "external_updated_at": 1392036272, "custom_attributes": { "order_number": "ORDER-12345", "total_amount": 99.99 }Now the documentation does not say anything about how to link the Order to People.Fin says, you need to put a customer property inside custom_attributes, fi

We are new integrating Intercom with our systems. Right now having problems with location object... Where can i get some list of valid (countries, regions, cities) in order to map correcctly?

Hi everyone 👋We’re building a Canvas App Sheet for a live Intercom Messenger integration. Our goal is to let logged-in users view and update profile data — specifically values like First Name, Last Name, Primary Chapter, and SpotMe_Id — all of which are being passed into Intercom via a custom embed snippet.✅ These custom attributes show correctly on the user profile inside Intercom.🚫 But when the Canvas App loads via /sheet, the contact.custom_attributes payload does not include these fields (we checked with full logging of req.body). We tried accessing them via: jsconst contact = req.body.contact || req.body.context?.contact || {}; const attrs = contact.custom_attributes || {}; Even though we can see the correct attributes in Intercom, attrs only contains empty/default values like Member, or Global.We’ve verified: intercomSettings in the embed includes the attributes (they appear in the user profile) We’re using open_mode: "conversation" in our Canvas App button App is launched

We’re looking for a way to mass import our knowledge base articles from Zoho Desk into Intercom. What’s the best way to accomplish this? 

Hello everyone, I encountered a strange issue while testing user flows for a mobile video editor, similar to the capcut apk tools I've been using. sometimes it takes a long time for the Intercom widget to appear when the app loads larger media files, or it appears after everything else has finished loading.Has anyone else observed that Intercom behaves strangely when handling large files or a lot of ui changes? Is there a better way to queue events to ensure smooth operation, or do you think this is just a performance hiccup? I'd be interested in learning how others have dealt with it.

Hi Support,We are investigating an issue with our Salesforce integration where data is not persisting for new users, specifically regarding a race condition between the Salesforce Sync and the JS SDK boot method.Our Setup: Integration: Official Salesforce App. Matching Logic: We currently match on Email. User Creation: Users are created via the JS SDK (Intercom('boot')) when they first log into our portal. The Scenario: Time T1: A Contact is updated in Salesforce. A mapped custom field (salesforce_resource_access) is populated. Time T1 + 5 min: The Salesforce Integration triggers a sync. Note: At this moment, no User exists in Intercom with this email address. Time T2 (Days later): The user logs into our web portal. Time T2 + 1 sec: We call Intercom('boot', { email: ... }). The user is created in Intercom. The Issue:When the user is created via the JS SDK at Time T2, the salesforce_resource_access field is blank. The data synced at Time T1 is lost.Our Questions: "Create N