Audience Builder: User Guide

Audience Builder is an AI-powered list-building tool on Civis Platform that lets you describe the people you want to reach in plain language — no SQL required. Type something like 'women over 45 in Georgia who have donated before' and the tool translates that into a filtered list you can preview, refine, and export.

Audience Builder is accessible from the Civis AI menu in the top navigation bar.

Export options include CSV, Redshift table, VAN/VoteBuilder, Meta/Facebook, Google Sheets, and Salesforce. Available export destinations depend on what your organization has configured.

Getting started

Before you begin

Before you can build an audience, your organization needs to have at least one dataset configured for Audience Builder. This requires an admin to connect a table, ensure it has descriptive column labels and a unique ID column, and grant you access to it. If you don't see any datasets available or encounter a permission error, reach out to your admin or contact support@civisanalytics.com.

Creating a new audience

Click New Audience — either from the top of the Audience List or in the left sidebar. If your organization has multiple datasets configured, you'll be asked to pick one. Your new audience opens immediately in the chat interface.

The layout

The screen is split into two panels:

• Left panel (Chat): Where you describe and refine your audience by typing.
• Right panel (Canvas): Opens when you click one of the panel buttons — Charts, Filters, Sample Records, Data Dictionary, or Export.

You can drag the handle between the panels to resize them, or press Escape to close the right panel.

Building your audience

Describing your audience

Type a description of who you want in the chat input at the bottom. Be as specific or as broad as you like — the AI translates your words into filters.

Examples of things you can type:

• "Women in Ohio between 35 and 55"
• "People who have donated in the last two years"
• "Registered Democrats in rural counties"
• "Everyone — remove all filters"

If you're not sure what attributes are available in your dataset, click the Data Dictionary button in the right panel to browse all available columns and their descriptions before you start.

The chat will respond with the filters it applied and how many people match. If the number doesn't look right, just say so: "That's too many — make them younger" or "Only include homeowners."

Refining your audience

You can keep refining in the same conversation:

• "Add people in Kentucky too"
• "Narrow it down to the top 500"
• "Switch to a random sample instead"
• "Start over"

Each message updates the audience. The filters always reflect the latest state — you can review them at any time by clicking the Filters button.

If results look unexpected, opening the Filters panel is the best first step — it shows exactly what criteria were applied so you can identify where to refine.

Editing filters manually

Click the Filters button to open the Filters panel. You can review each active filter and edit values directly without going through chat. Scroll down in that panel to find the Settings section.

This is useful when you want to make a precise adjustment — such as changing an exact age range or score threshold — without retyping a full description in chat.

Settings

The Settings section (in the Filters panel or accessible via chat) has two options:

Max audience size Sets a cap on how many records are included. Leave it blank to include everyone who matches your filters. When a limit is set, the "Sort by" option appears.

Sort by Controls which records are included when the list is capped. Options depend on what your admin has configured, but you'll always have "Randomized order" available. If score columns are configured (e.g., a donor score or engagement score), you can sort by those to get the highest-value contacts.

If you don't see a score column you expected, contact your admin or reach out to support@civisanalytics.com to have it configured.

You can also change these settings through chat: "Give me the top 1,000 sorted by donor score" or "Give me a random sample of 500."

Note that your Max audience size setting persists as you continue refining in chat. If you want to see the full count of records matching your filters, clear the size limit before exporting.

Viewing your results

Charts

Click the Charts button to open the Attributes Comparison panel. When no filters are active, charts show the full dataset distribution. When filters are active, your audience's distribution is shown alongside the full dataset (labeled "All People") so you can see how your audience differs.

The list size limit applies to charts — if you've set a cap of 500, charts reflect those 500 people, not the full filter-match count.

The order charts appear reflects the column order in your dataset. If you're looking for a specific chart and can't find it, try scrolling down in the Charts panel.

Sample Records

Click the Sample Records button to browse a preview of actual records in your audience. This is useful for a quick sanity check before exporting.

The preview shows a random sample of records from your current audience, not the full list.

Data Dictionary

Click the Data Dictionary button to see all available columns in the dataset, their descriptions, and sample values. Useful if you're not sure what attributes are available to filter on, or if a filter returned unexpected results and you want to review how a column is described.

Exporting

Click the Export button to open the Export panel. Export options depend on what your admin has enabled for your organization.

Available export types

• CSV: Downloads a gzipped CSV file to your computer.
• Table: Writes the audience to a Redshift table.
• VAN: Pushes a Saved List directly to VAN/VoteBuilder. The list name in VAN will match the audience name you set in Audience Builder. Requires admin configuration.
• Meta/Facebook: Pushes a Custom Audience to Meta (requires admin configuration).
• Google Sheets: Exports to a Google Sheet. You can use your own Google account connected in Civis, or an org-wide account your admin has configured. Options include creating a new spreadsheet, adding a new sheet to an existing one, overwriting a sheet, or appending to one.
• Salesforce: Pushes records to a Salesforce object (requires admin configuration).

After an export finishes, it appears in the export history with its status. For Google Sheets exports, an Open Google Sheet button appears so you can jump directly to the result.

If an export fails, its status will appear in the export history. Reach out to support@civisanalytics.com with the audience name and export type and we can help troubleshoot.

Choosing which columns to export

Before exporting, you can choose which columns to include. By default all available columns are selected. Column selection happens before you confirm the export, so review your selection before clicking confirm.

Managing audiences

Renaming an audience

Click the audience name at the top of the chat to rename it inline. The AI will also suggest a name based on your filters when you first build one — you can keep it or change it.

Copying an audience

To iterate on an existing audience without changing the original, click the copy button on the audience in the Audience List. The copy opens as a new audience with the same filters and chat history, which you can then refine independently.

The Audience List

The home screen shows all audiences in your organization. You can filter by owner (defaulting to your own audiences) to find your work or browse others'.

Ownership rules:

• Only the owner of an audience can edit filters, chat, or export.
• Other members of your organization can view an audience (filters, chat history, sample records) but cannot make changes.
• Audiences are owned by the person who created them. If you need to modify an audience created by a colleague, ask them to make the change or have them create a copy and transfer it to you by contacting support@civisanalytics.com.

Archiving

You can archive audiences you no longer need. Archived audiences are hidden from the default list view but can be restored. To view archived audiences, filter the Audience List by archived status. To restore one, open it and click Restore.

Tips

• Be conversational. You don't need to use technical terms. "People who are older" works just as well as specifying an age range — the AI will apply something reasonable and tell you what it chose.
• Explore the Data Dictionary first. If you're new to a dataset or not sure what attributes are available, open the Data Dictionary panel before you start typing. It shows every column, its description, and sample values — knowing what's there helps you write better queries and avoid unexpected results.
• Check the count. The chat always tells you how many records match. If the number seems off, describe the problem: "That seems like too few — why?" and the AI will explain the filters.
• Verify the filters. AI translates your description into filter criteria. The actual data lookup is done by standard (non-AI) software, so the counts are accurate — but it's worth opening the Filters panel to confirm the criteria match your intent.
• Check your dataset description. If queries keep returning unexpected results, the issue may be with how your dataset is described rather than how you typed your query. Contact support@civisanalytics.com and we can review the table setup with your data team.
• Try the Charts panel. It's a fast way to see whether your audience skews the way you expect before you export.
• Need help? Contact Civis Support at support@civisanalytics.com.