How to see which Intercom conversations lead to conversions

You see which Intercom conversations lead to conversions by linking two things Intercom stores separately: what each conversation was about, and what the person behind it did next. Every conversation in Intercom belongs to a contact. That contact record carries the evidence of conversion, so the work is to read the conversations, sort them into meaningful groups, and then count how many people in each group converted.

Intercom gives you part of this on its own. Its reports cover conversation volume, response and resolution times, customer satisfaction ratings, and Fin's resolution rate, and it can group conversations into topics. Those numbers tell you how well your team and Fin handled the conversations. They stop short of the outcome, because a conversation report doesn't look at whether the visitor later signed up or became a paying customer. Intercom does let you set a goal on outbound messages, such as a Series or a Post, but that goal measures the message you sent, not the conversations people started with you.

The outcome is usually sitting in the contact record. When a visitor chats before signing up, Intercom stores them as a lead. When they later sign up and your app identifies them to Intercom, Intercom turns that lead into a user, and the conversation history comes along. Your app may also send a plan, a sign-up date, or a paid status as custom attributes, and product events such as a completed setup step or a started subscription. Company records can carry a plan and monthly spend.

A good answer to this question looks like a ranked list of conversation topics, each with how many people asked about it and how many of them converted. It also needs one honest correction. People who open a chat are often already interested, so a high conversion rate in one topic only means something when you compare it with the other topics and with people who never chatted. Kite does this work for you inside Intercom, and it tells you what to change based on what it finds.

Start by asking Kite which conversations lead to conversions, in the Kite web app or in Slack:

"Which kinds of Intercom conversations from the last 90 days ended with the person becoming a paying customer?"

You can also narrow the question to one group of visitors or one kind of conversation:

"Look at the conversations that started on our pricing page this quarter and tell me which questions come before people upgrade and which come before they go quiet."

Kite is an AI marketer. It researches your business, builds a growth strategy, and ships the work that brings you new customers. By the time you ask, it already knows what you sell and how people buy from you, so it can propose what should count as a conversion, such as a trial sign-up, a paid plan, or a booked demo, along with the period to study. Confirm those or tell Kite what you count instead, and it starts.

You'll need to connect your Intercom workspace through OAuth. An Intercom admin signs in and approves access, and Kite can then read the conversations, contacts, and companies that admin's permissions allow. Reading and analyzing your conversations is routine work, so Kite does it and reports back. Before it adds tags or attributes to your Intercom workspace, it shows you what it plans to add.

Kite first decides which change on a contact means that person converted, because Intercom doesn't have one built-in conversion field. It reads the data attributes defined in your workspace for contacts, companies, and conversations, so it can see what your app actually sends to Intercom. Depending on your setup, the conversion might be any of the following:

• a lead becoming a user, which usually means the person signed up
• a sign-up date appearing on the contact
• a custom attribute, such as a plan name or a paid status, changing to a paying value
• a product event, such as a started subscription or a completed setup step
• a company's plan or monthly spend moving above zero

Kite proposes the signal that matches what you told it you care about, and it tells you if your app isn't sending that signal to Intercom yet. When a conversion only exists in another system, such as your billing tool or CRM, Kite says so and can follow the contacts there through that connection instead.

Kite gathers the conversations from the period you chose and reads each one in full. It searches your conversations by creation date, and it walks through every part of each thread, including the visitor's messages, your team's replies, and Fin's answers. For each conversation, it records:

• the page the conversation started on
• when it started and whether it's closed
• how long the first reply took
• any rating the visitor left
• the tags already on it
• the contact behind it

Kite leaves out conversations that can't tell you anything about buying, such as conversations from existing customers asking for routine support, unless you asked about upgrades. It also notes conversations that have no contact it can follow, so you know how much of your chat volume the answer covers.

Kite sorts the conversations into topics based on what the visitor was trying to learn, using your customers' own words rather than a fixed list. Typical topics include:

• pricing and plans
• a specific integration
• security and compliance
• how you compare with an alternative
• help getting started

A single conversation can touch more than one topic, so Kite records the main reason the person reached out and notes any second question.

Reading the full thread matters here. A conversation that opens with "Quick question" and turns into a long discussion of single sign-on belongs in the security group, and a search for keywords alone would miss that. Kite shows you the topic list with a few real example questions in each before it goes further, so you can merge or rename groups to match how your team thinks about them.

Kite follows each conversation's contact to see whether the conversion happened after the conversation and within a sensible window. It reads the contact's role, sign-up date, custom attributes, tags, segments, and company, and it reads the product events your app has recorded on that contact. Intercom keeps those events available for 90 days, so for older conversations Kite relies on the contact's attributes and dates instead.

The order of events matters as much as the result. A person who was already a paying user when they chatted didn't convert because of the conversation, so Kite separates conversations that came before a conversion from conversations that came after it. It also picks a conversion window that suits how people buy from you, such as two weeks for a self-serve trial or a few months for a sales-led deal.

Leads who chatted from more than one browser or device sometimes end up as two contacts in Intercom, one lead and one user. When Kite finds a lead and a user with the same email address, it treats them as one person for the analysis. With your agreement, it can also merge the lead into the user in Intercom, so the conversation history sits on the right record from then on.

Kite compares the topics with each other and with a baseline, because the raw conversion rate of people who chat mostly measures how interested they already were. It compares each topic's conversion rate with the rate across all conversations, and, where your workspace holds users who signed up without ever chatting, with that group as well. It also compares conversations that started on the same page, so pricing-page questions are measured against other pricing-page questions.

Inside each topic, Kite looks for the differences you can act on. It compares closed conversations with ones that were left open, quick first replies with slow ones, and answers from Fin with answers from your team. When a topic has too few conversations to support a conclusion, Kite says so plainly instead of ranking it.

Once you agree with the topics, Kite tags each conversation in Intercom with its topic, so the analysis keeps working after the report is finished. If you'd like a single field instead of tags, Kite can create a custom attribute for the topic. Your team can then filter the Intercom inbox and reports by topic, and Intercom's own reports start showing how volume and reply times differ between topics.

Kite can also tag the contacts whose conversations came before a conversion. That gives you a segment of people who chatted and bought, which is useful for finding customer stories and for targeting the messages you send from Intercom.

When the analysis is done, Kite reports in the conversation where you asked, starting with the answer. The report names the topics that most often come before a conversion and the topics that most often come before people go quiet, with the numbers behind each one. It also says how many conversations it could follow to a contact and which conversion signal it used.

Kite then recommends what to do and does the work you approve. A question that keeps coming before people leave usually means your website doesn't answer it, so Kite might propose adding the answer to the page where those chats start, writing a help article that Fin can use, or changing your sales team's priorities toward the topics that convert. It can make the website change on a site it hosts, then measure whether conversions rise and whether fewer visitors need to ask. If you want the report every month, ask Kite to put it on a schedule, and each run arrives in the same place with that period's numbers.

When Kite finishes, you have a report that connects your Intercom conversations to the conversions that followed, plus topic tags in Intercom that keep the connection visible to your team.

For a longer report, Kite can publish it as a findings page or a dashboard at a stable link that refreshes its data when you open it.