If you want Fin AI Agent and Copilot to tailor answers based on customer segments and data, you can create and apply audiences.

Fin audiences are reusable customer segments that you define using People and Company data. You can then apply these audiences across various Fin features to deliver personalized experiences.

Key principles of how audiences work:

To define new Fin audiences, go to Fin AI Agent > Fin settings > Audiences. Click on New audience and give it a name.

Now set the audience targeting rules that define that audience using data attributes. You'll then be able to see a preview of the audience it matches (the preview is not available if conversation attributes are selected).

Currently this supports:

When you create an audience based on a Brand, you can also give Fin a unique name and avatar. Learn more about setting up Fin identities for a branded AI experience.

Once this audience is saved, you can then reuse it across Fin brand identities, content, guidance, data connectors, tasks, and workflows.

Once your Fin audience is created, you'll be able to select this audience for your content, guidance, data connectors, tasks, and workflows so Fin provides tailored support to that group of customers.

By applying a Fin audience to your content, you can ensure that the Fin uses only the appropriate sources for a specific group of customers. This allows Fin to deliver tailored and accurate support.

To apply a Fin audience to your content, select the content items in Knowledge > Content, then click on More actions and select Change audience.

By applying a Fin audience to guidance, you can ensure Fin's tone and policies match the customer group it's talking to.

By applying a Fin audience to data connectors, you can choose which external data sources, like your store or CRM, Fin can use for each customer group.

By applying a Fin audience to tasks, you can choose which tasks are triggered for each customer group.

By applying a Fin audience to workflows, you can choose which workflows are triggered for each customer group.

On the Workflows overview page, you can filter by Fin Audiences — making it simple to see which workflows use a particular audience.

It's possible to select multiple audiences when targeting with Fin, in this case they are treated like an “OR” rule. For example, if Audience A and Audience B were selected for the same piece of content, Fin would be able to use that content for any user that matches Audience A, or Audience B, or both.

You can still create a Custom audience for tasks and workflows using attributes and predicates if you need a specific audience set up for one particular task/workflow. There's also an Everyone option that can be selected to include all users, visitors, and leads.

Use your Fin audiences to make AI-powered suggestions more accurate and relevant. By segmenting suggestions, Fin analyzes conversations belonging to specific customer groups to help you identify automation gaps. This allows you to tailor your support content and data to the precise needs of each audience. For a step-by-step guide, see our full article on how to use AI suggestions by audience.

You can review your audience configuration before your customers interact with Fin. This allows you to see exactly where an audience is being used.

Visit Fin AI Agent > Fin settings > Audiences and select an audience to see a summary of all items where Fin is using this audience.

To find content that matches a specific audience, use the Audience filter at the top of your content list in Fin AI Agent > Train > Content.

To see guidance that matches a specific audience, use the Audience filter at the top of your guidance list in Fin AI Agent > Train > Guidance.

You can use Batch Testing to check Fin's answers for a specific audience.

You can also use Fin previews to quickly check Fin's answers or behavior in the context of adding new content, guidance, etc.

When using Fin, it's important to understand the two different types of audience targeting you can apply to public articles. These settings work together to control who can see an article in your Help Center and if Fin can use an article when responding to a customer.

Fin will only use an article to answer a customer’s question if both conditions are true:

Example

Your business has two plans: Basic and Pro.

If a Basic Plan customer asks Fin about those features, Fin won’t use the article because the customer doesn’t meet the Help Center audience rules.

When Copilot provides an answer, it can also use the past 4 months of conversations. But sometimes allowing Copilot to use all of your previous conversations can result in inaccuracies or inconsistencies in answers, because the answer didn’t take into account the customer context.

If you've enabled conversations for Copilot, this is where you can select attributes for Fin to match past conversations with.

Target audiences are applied differently for past conversations because you potentially have thousands of past conversations that Copilot can utilize, and they're not shown in your Knowledge Hub, so you don’t have the ability to define a unique audience for each one.

Instead, you can select the attributes that you’d like Copilot to use when matching past conversations with the current conversation that you’re working on.

Go to Knowledge > Sources and open the For AI Copilot tab. Then scroll down to "Conversations" and click Manage.

Under “Conversations with matching attributes” click + Add attribute.

Now select the attribute(s) that you’d like to use. In this example, we'll select “Product Area” because we want Copilot to use past conversations with the same product area as the current conversation we’re working on.

You can also combine "And" / "Or" rules when matching multiple attributes. For example, if you want Copilot to use past conversations with the same product area and plan.

Remember to click Save once you've finished selecting attributes.

When you ask Copilot a question it will now check to see if the customer conversation has that attribute(s) defined. If it does, it will only answer using past conversations with the matching attribute(s).