What Actually Syncs When You Connect Intercom to Salesforce or HubSpot

Business owners running Intercom often assume that connecting Salesforce or HubSpot will finally get sales and support looking at the same picture. It's a reasonable hope. But the two integrations were built for different jobs, and neither one does everything people expect. If you're about to wire your CRM into Intercom, or you already did and are wondering why some fields never show up, it helps to know exactly what moves between the two systems and what stays put.

What the Salesforce Integration Actually Does

Intercom's native Salesforce integration is built around leads and contacts moving in both directions. When someone becomes a lead or contact in Intercom, that record can be pushed into Salesforce, and updates made on the Salesforce side can flow back into Intercom too. Custom fields sync as well, as long as the data types line up between the two systems, so a text field in one has to map to a text field in the other rather than something like a picklist.

Ownership carries over automatically. If a lead or account has an owner assigned in Salesforce, that owner maps to the matching teammate in Intercom, which is what makes it possible to route a conversation to the right rep without someone manually checking Salesforce first. Finished conversations also generate a task on the Salesforce side, logged against the relevant record, so a sales rep can see that a prospect talked to support without switching tools.

This integration is only available on Intercom's higher-tier plans, and it needs a Salesforce account with API access enabled, which usually means Professional edition with an API add-on or higher. The Salesforce user connecting the integration needs to be a real seat, not an API-only user, with read and write permissions on leads, contacts, accounts, and tasks.

There are real limits worth knowing before you set expectations. Tags and event data don't map across, formula fields only update when something else on the record changes, and fast-moving attributes like last-seen timestamps aren't part of the sync at all. Data also isn't instantaneous. There's typically a short delay, so don't expect a Salesforce update to appear in an Intercom conversation the second it happens.

Why HubSpot Works Differently

The default Intercom-HubSpot app looks similar on the surface but behaves quite differently underneath. It only sends data one direction, from Intercom into HubSpot. Nothing flows back the other way through that app. If a rep updates a deal stage in HubSpot, Intercom has no idea it happened.

What does go across is fairly narrow: contact details like name, email, and phone number, qualification data collected through Intercom's bots or forms, and conversation transcripts. Transcripts sync in a slightly uneven way too. Messages from the customer sync in close to real time, but a team member's replies don't show up in HubSpot until the conversation is marked closed.

The fields that sync are fixed. You can't add your own custom mapping to this particular app the way you can with Salesforce, and every new contact created this way gets dropped into HubSpot's "Lead" lifecycle stage automatically, whether that's accurate or not. If your team needs data to flow both ways, Intercom points people toward a separate two-way sync app built specifically for that, rather than the basic HubSpot card.

Where Fin AI Actually Gets Its Answers

Here's the part that trips people up the most. Neither the Salesforce nor the HubSpot integration is what powers Fin AI's answers about a customer's order, account status, or subscription. Those integrations exist to keep sales and marketing informed about what's happening in support, not to feed live data into the AI agent's replies.

The mechanism that lets Fin pull in that kind of real-time, per-customer information is called a data connector. A data connector is a direct link to an API, whether that's an internal order system, a billing platform, or a third-party app, that Fin can query while it's actively answering a question. Instead of giving a generic answer about shipping policy, Fin can look up the specific order tied to that customer and answer with their actual delivery date.

Setting these up well matters more than people expect. Fin needs a clear, specific description of when to use each connector, written with concrete examples of the kinds of questions that should trigger it. Vague instructions lead to Fin either not using the connector when it should, or using it for the wrong situation. It also works best when the API call identifies the customer by something unambiguous, like an account ID or email address, rather than a loosely defined parameter that could match more than one record. Large or messy API responses should be trimmed down before they reach Fin, since a huge nested payload is more likely to produce a confused or inaccurate answer than a clean, filtered one.

Common Mistakes Worth Avoiding

The most frequent mistake is assuming a CRM integration equals real-time, bidirectional sync by default. With HubSpot's basic app, it explicitly isn't. Assuming otherwise means someone eventually makes a decision based on stale data without realizing it.

Another common one is trying to sync tags, custom events, or frequently changing attributes through Salesforce and being surprised when they don't appear. Those types of data were never part of what the integration handles, so it's worth checking Intercom's current list of supported field types before promising your team a specific piece of data will show up automatically.

The third mistake is treating the CRM integration and Fin's data connectors as the same thing, then wondering why Fin can't answer questions about order status even after Salesforce is connected. They solve different problems and usually need to be set up separately.

Deciding What You Actually Need

If the goal is keeping your sales team aware of support conversations and routing leads correctly, either the Salesforce integration or the basic HubSpot app can get you there, with Salesforce offering more depth if two-way sync matters to you. If the goal is having Fin answer detailed, account-specific questions on its own, that's a data connector project, and it's worth scoping separately from whatever CRM sync you already have running.

Getting this stack set up correctly, and getting Fin's connectors written so they actually trigger at the right moments, is exactly the kind of tuning work I help businesses with. If you're trying to connect the dots between your CRM and your Fin AI setup and want a second pair of eyes on it, feel free to reach out.

Cover photo by Annie Spratt on Unsplash.