Conversation Intelligence Integrations: How It Connects to Salesforce, HubSpot and Zoom

Conversation intelligence integrations connect the tool that hears your customer conversations to the CRM where your team plans, forecasts and reports. For a lot of RevOps leads and sales managers, those two systems still do not talk. The call tool knows the prospect pushed back on price and named a competitor. The CRM shows a logged call with an empty notes field.

So reps retype notes after every call, or the insight stays in a separate tab nobody opens during pipeline review. That admin adds up: in Salesforce's State of Sales report (6th edition, 2024, 5,500 sales professionals), reps said 70% of their time goes to non-selling tasks. A good conversation intelligence CRM integration takes part of that work off their plate. This guide covers what data moves, three ways to set it up, what to expect from Salesforce, HubSpot and Zoom, and what to check before you go live.

What Are Conversation Intelligence Integrations?

A conversation intelligence integration connects your call and video tools (a phone system, a dialer, Zoom) with your CRM (Salesforce, HubSpot) so conversations are recorded, transcribed and analyzed, and the results land on the right contact, company or deal.

If you need the basics first, read our guide to conversation intelligence software: definition and benefits. This article focuses on the connection between the two systems and what it takes to make it work.

Here is what typically moves from the conversation tool into the CRM:

Integrations also differ in direction. In a one-way integration, the conversation intelligence tool writes to the CRM. In a two-way integration, the CRM also sends context back, such as deal stage, owner and amount. That context lets the tool group calls by deal and flag risk on late-stage opportunities, which a one-way sync cannot do.

How the Data Flows: From Call to CRM Record

Every integration has three layers:

• The source: Where the conversation happens. This is a cloud phone system, a dialer, Zoom, Microsoft Teams or Google Meet.
• The conversation intelligence engine: The layer that turns audio into a call transcription, then analyzes it for summaries, topics, sentiment and talk time.
• The destination: Where the results go. This is usually the CRM, and sometimes also Slack, a BI tool or a coaching platform.

When the source and the engine are the same product, for example a phone system with built-in AI, there is one fewer connection to maintain. When they are separate, a recorder or an import step has to run before any analysis starts, and that step is one more thing that can break.

Matching the Call to the Right Record

This is the step that fails most often. The tool has to work out who the conversation was with, then attach it to the right contact and the right deal. For phone calls, it matches by phone number. For video meetings, it matches by the participants' email addresses from the calendar invite.

Matching to the contact is usually easy. Matching to the deal is harder: if a contact has two open deals, or none, the tool has to guess or skip the call. When teams say the integration does not work, unmatched calls are the usual cause. Typical triggers are phone numbers saved in different formats, prospects calling from a mobile number that is not in the CRM, and meetings booked from a personal calendar. Consistent number formats and a clear rule for which deal gets the call (the most recent open deal, for example) fix most of these cases.

Real-Time or Batch?

Real-time sync pushes data while the call is running or seconds after it ends. It helps with live alerts, such as telling a manager when a competitor comes up, but it costs more to build and run. Batch sync processes the call after it ends and writes results within minutes. Most sales teams only need the summary and next steps on the deal before the rep's next call, so after-call processing covers the common use cases.

APIs and Webhooks for Custom Builds

Teams that build their own pipeline work with events: call ended, recording ready, transcript ready, summary ready. Each event triggers a webhook that tells your system to fetch the data and write it where you need it. A conversation intelligence API lets you send transcripts and analysis into a data warehouse, an internal tool or a CRM object that the standard connector does not support.

For example, Twilio Conversation Intelligence (labeled "classic" in Twilio's current docs) offers a REST API for transcripts and language operators, plus webhooks that send results to your application while the analysis runs.

Build your own pipeline when your needs are very specific and you have engineers to maintain it. Buy a connector when your needs are standard, such as a summary on the deal and a transcript on the activity, because every custom integration needs an owner when an API changes.

Three Ways to Integrate Conversation Intelligence With Your CRM

Before you compare vendors, decide which kind of integration you want. Each approach puts the analysis in a different place, and that decides what you pay, what you maintain and which conversations get covered.

CRM-Native Conversation Intelligence

Native tools live inside the CRM, so insights sit next to the deal without a second login. Salesforce, HubSpot and Microsoft all offer one. The conversation intelligence Dynamics 365 Sales includes, for example, works with Teams calls and meetings and with third-party telephony, as long as recordings are in stereo (two-channel). The catch is licensing: native tools usually need a higher-tier plan or an add-on, and they only analyze recordings from sources the CRM supports.

Standalone Conversation Intelligence With Connectors

Standalone platforms such as Gong and Jiminny offer the deepest analysis and coaching features, and they connect to several CRMs. The trade-off is one more tool in the stack. Salesforce Ben argues that non-native AI notetakers "create another data silo" because conversation data is stored somewhere else. That concern is fair, but how much it applies depends on the connector. A tool that writes summaries and scores into CRM fields becomes part of your CRM data. A tool that only links out to its own app is a silo.

Telephony-Native Conversation Intelligence

Here, the phone system records and analyzes the calls itself, then pushes results to the CRM. There is no recorder to connect, and every call made through the system is covered. The gap is video: if half your selling happens on Zoom, you need another tool for those meetings.

A related pattern: some phone systems only pass recordings into the CRM's own conversation intelligence. In that setup, the CRM seat and plan still decide what analysis each rep gets, so check both vendors' requirements.

Contact center suites: Genesys, NICE and Amazon Connect build conversation intelligence into the suite. Salesforce also lists Amazon Connect as a supported recording source for its native tool.

Quick decision rule: Video-first teams should pick a CRM-native or standalone tool that records meetings. Phone-first teams should look at telephony-native tools, and teams that run everything in one CRM should start with that CRM's native option.

Salesforce, HubSpot and Zoom: What to Know for Each

Each platform handles conversation data differently. Below is what you get natively, what you need, how data reaches records and where integrations usually break. Requirements come from each vendor's documentation as of November 2026. They change often, so check the linked pages before you buy.

Salesforce Conversation Intelligence

Salesforce's native tool was called Einstein Conversation Insights and has been rebranded to Conversation Intelligence, which explains the mixed names in search results. According to Salesforce's Trailhead requirements, it is included in Performance and Unlimited Editions and costs extra in Enterprise Edition. It analyzes recordings from Zoom Meetings, Salesforce's Sales Dialer and listed telephony partners such as Dialpad, RingCentral, Aircall and Amazon Connect. Other partners can send recordings through an API.

When you compare conversation intelligence Salesforce options, you will find both native features and third-party apps on AppExchange, such as Authenticx's Lightning app. Either way, the useful outputs on an opportunity are the same: a call summary, objections, competitor mentions and next steps. Many teams also send these insights to Slack so deal teams see them without opening the CRM.

Common gap: Insights stay in a separate tab or related list and never reach fields you can report on. Map the outputs you care about, such as next step, competitor mentioned and sentiment, to custom fields on the opportunity so they appear in pipeline reports and forecast reviews.

HubSpot Conversation Intelligence

HubSpot conversation intelligence transcribes and analyzes call recordings for users with an assigned Sales Hub or Service Hub Professional or Enterprise seat, according to HubSpot's knowledge base. Calls made with HubSpot's own calling tool are covered, and so are calls from the select third-party providers HubSpot supports. Through HubSpot's Zoom integration, Zoom cloud recordings appear on contact timelines, as long as cloud recording and audio transcripts are turned on in Zoom.

The core HubSpot conversation intelligence features are transcripts, call summaries and tracked terms. On the AI side, Breeze call summaries cover the purpose of the call, key discussion points, decisions, sentiment and next steps. To use HubSpot conversation intelligence AI summaries, an admin turns on Breeze Assistant and call recording.

For HubSpot conversation intelligence coaching, managers leave timestamped, threaded comments on a recording, @-mention reps and add calls to coaching playlists. Tracked terms (Enterprise only, set up by Super Admins) work as a simple scorecard. To build a HubSpot conversation intelligence MEDDIC view, add tracked terms for each element of MEDDIC (Metrics, Economic buyer, Decision criteria, Decision process, Identify pain, Champion), such as "budget", "sign-off" and "timeline", then filter calls where they did or did not come up.

Common gap: Third-party dialers feed HubSpot's native tool only through listed integrations, and only users with the right seat get transcripts. If you are looking for a conversation intelligence HubSpot setup that works with your existing dialer, check both lists before you buy. CloudTalk's HubSpot integration is in the HubSpot marketplace on CloudTalk's Essential and Expert plans. It adds calling inside HubSpot, two-way contact sync, call outcomes from tags, and, with CloudTalk's Conversation Intelligence add-on, AI insights such as call summaries, sentiment and topics in HubSpot engagement notes. CloudTalk pushes these insights to HubSpot, Pipedrive and Zoho today.

Zoom Conversation Intelligence

Zoom conversation intelligence means two different things. First, Zoom as a source: cloud recordings and transcripts feed CRMs and standalone tools, and Salesforce, HubSpot, Gong and Jiminny all import them. Second, Zoom's own product: the tool once sold as Zoom IQ for Sales is now Zoom Revenue Accelerator, and Zoom lists it under conversation intelligence on its site.

As a Zoom conversation intelligence platform, Revenue Accelerator offers bi-directional sync with Salesforce, HubSpot and Microsoft Dynamics 365, according to Zoom's integration docs. It pulls deal and account data from the CRM, refreshed every four hours, and pushes the insights an admin selects back as activities: summaries, sentiment and engagement scores, next steps, competitors mentioned and a link to the conversation.

Zoom native features, conversation intelligence included, only cover meetings and calls that run on Zoom. Before you add any Zoom conversation intelligence software, check which Zoom plan and add-ons it needs, because that decides the real cost.

Common gap: Cloud recording must be turned on, because local recordings do not sync, and you need Zoom admin rights to connect the account. HubSpot's Zoom integration also needs audio transcripts turned on.

Gong and Other Standalone Platforms

A Gong conversation intelligence integration with Salesforce works in both directions. Gong reads CRM data to attach each call to the right account, opportunity, contact or lead, and its managed package writes conversation records, and optionally activities, back into Salesforce. Gong also exports data to HubSpot. If you are weighing a standalone platform against a phone system with built-in analysis, see our CloudTalk vs Gong comparison.

How the main tools compare:

What to Check Before You Integrate, and Common Problems

A Pre-Integration Checklist

• Objects and fields: Ask which CRM objects (contact, company, deal, activity) and which fields the tool writes to.
• Reporting: Confirm you can build reports and dashboards on those fields, not only view insights in a tab.
• Sync direction: Check whether the sync is one-way or whether the CRM sends deal stage, owner and amount back.
• Call matching: Ask how the tool picks a deal when a contact has more than one open deal.
• Plans and seats: List the CRM plan and the seat each rep needs to see transcripts and summaries.
• Permissions and SSO: Find out who needs admin rights to connect it and whether it supports your single sign-on.
• Languages: Make sure transcription covers every language your team sells in.
• Cost: Add up the full cost, including CRM upgrades, add-ons and per-seat fees.
• Privacy: Check where data is stored, how long it is kept and how you collect recording consent under GDPR and, in healthcare, HIPAA. See our guides on whether call recording is legal and HIPAA call center requirements.
• Small teams: Use native features or prebuilt connectors and skip custom builds, because nobody will have time to maintain them.
• Enterprises: Check SSO, user provisioning with SCIM (an automated way to add and remove users), central logging and export to your data lake.

CloudTalk's integrations guide lists the CRM and sales tools CloudTalk connects to, including Salesforce, HubSpot, Pipedrive, Zoho, Copper, Outreach, Salesloft, ActiveCampaign and Microsoft Dynamics 365 Sales.

Common Problems and Fixes

A 30-Day Start

1. Week 1: Connect the integration for one team and one workflow, for example logging AI summaries and next steps on deals.
2. Weeks 2 and 3: Let the team work as usual and collect questions from reps and managers.
3. Week 4: Check 20 calls by hand: is each one on the right deal, is the summary accurate, and did any manual notes get overwritten? Fix what you find, then roll out to the next team.

What Integration Is Worth: Benefits and KPIs

The value shows up in three places. Track one or two numbers for each so you know whether the integration pays off.

Automated Data Entry

AI call summaries and tags write the notes so reps do not have to. Track the time reps spend on notes after each call and the share of calls with a completed next-step field.

Deal Coaching and Risk Flags

When insights sit on the deal, managers can review the right calls instead of random ones. Track how many calls managers review each week and how many at-risk deals get flagged before they slip. Coaching also gets easier when highlights and clips go straight to your enablement tool, so new reps learn from real calls.

One Workflow Instead of Many Tabs

Reps work in the CRM, and the insights are already there. Track how many tools reps open per call and how many reps actually use the insights each week.

Make Conversation Data Work Inside Your CRM

Conversation intelligence integrations move recordings, transcripts, summaries, topics and scores from your calls and meetings onto the right CRM records. You can get there three ways: a CRM-native tool, a standalone platform with connectors, or a phone system that analyzes calls itself. Salesforce and HubSpot tie their native tools to specific editions and seats, and Zoom works both as a source and as its own product.

Whichever route you choose, run the checklist first: know which fields get written, how calls match to deals and which seats you need. Then start with one team and check the data by hand. If your team sells mostly by phone, see how CloudTalk AI analyzes every call and logs the results for you.

Sources:

1. Salesforce: State of Sales, 6th edition (2024)
2. HubSpot Knowledge Base: Review call recordings and transcripts
3. Salesforce Trailhead: Conversation Intelligence requirements
4. Zoom: Revenue Accelerator integrations
5. Microsoft Learn: Conversation intelligence FAQs