Use conversation topics for the most consistent categorization of what your customers talk about, historically and into the future.

Let Intercom’s machine learning track and suggest conversation trends, then dive deep into exactly how your team handles each one. See customer satisfaction, response & resolution times, and filter or group other reports by topic.

A conversation topic is a way for you to automatically categorize every conversation in your workspace.

It’s a conversation property, which is automatically applied to matching conversations based on key phrases you define. Conversation topics allow you to deeply understand your conversations, and how your team performs when handling different topics.

Unlike conversation tags, and custom conversation data, topics can be applied retroactively to your existing conversations, and require no manual input from your customers or team.

To see a visual breakdown of the topics your customers chat about, and new suggestions, go to Reports > Conversation topics > Topics.

👆This page shows which topics generate the most conversations, with the higher volume topics represented by larger circles.

Intercom’s machine learning algorithm gives you a head start on topics to track, with suggestions presented under Reports > Conversation topics > Suggestions.

The Machine Learning algorithm that powers conversation topic suggestions typically needs about 5,000 total conversations to generate meaningful suggestions. The rule of thumb is the more conversations, the better the suggestions will be.

If your total conversation volume is lower than about 5000, you won’t have suggestions, however, when your total conversation volume exceeds this number, suggestions will start to show.

Eligible conversations for topic suggestion includes those which are:

Suggestions are potential groupings of conversations identified by machine learning that
share similar keyphrases. In this example, you can see the topic "Tasks", based on the keyphrases:

For a deeper understanding of how your customers feel about these topics, enable “Show customer satisfaction” and see the average customer satisfaction for each topic:

Next, switch to sorting by responsiveness instead of volume:

Now you’ll see your conversation topics on a scatter plot, with median time to close and median first response time on the Y and X axes.

This tells you what your customers are asking about, and what that means for your team.

For some quick efficiency gains, the conversation topics in this area are those with the quickest response time, and the quickest time to close.

Below the chart, each suggestion also shows how the topic is trending against the previous period in a number of areas:

This lets you zoom in on areas where your team’s performance could be improved, with things like:

To save a suggestion as a topic, select it from the list, and click + Add to your topics or select multiple and click Merge suggestions.

You’ll then see them in the list of saved topics, where you can also see the same metrics. 👌

To show the stats for a certain period of time just use the filter at the top of the page. You can also choose to exclude time spent in the Bot inbox from the data shown.

To start tracking a topic that isn’t in your conversation topic suggestions, go to Reports > Conversation topics > Topics and click New topic.

Then, enter all the relevant words and phrases your customers use to talk about this topic and separate them by pressing enter.

Optionally, you can also add keywords to exclude certain conversations:

For further refinement, you can choose to include conversations where teammates used the keywords, or just conversations where users used them.

Finally, you can add any other conditions based on your conversation data, like this example which filters for a certain “Product Area”:

You’ll immediately see conversations appearing so you can take a quick glance and ensure the right conversations matching the keywords and rules you’ve defined.

Finally, give your topic a title and save it.

This conversation topic will immediately be applied to all matching conversations, you’ll see it in the Inbox, and can utilize it as a property in your reports.

With topics tracked in Intercom, you can easily see how your team performs in certain areas at a glance, or you can open a topic to explore the specific conversations that match it. Just go to Reports > Conversation topics > Topics and click on a topic.

Here you’ll see a list of all the most recent conversations that match the topic, and you can open them up to see more.

If you find conversations that don’t match your topic as well as you’d hoped, you can always edit the topic to add or remove keywords.

As conversation topics are automatically applied to conversations (when they match the keywords and rules you’ve defined) you’ll see this reflected in the Inbox under "Conversation attributes".

You have full control here, so if you’d like to remove a specific topic from a conversation, just click the X on a topic. Or, click the + icon to add a topic manually.

Once you have defined conversation topics, you can use them to filter your reports.

Or use topics to filter and segment your very own custom reports.