Slack Bot for Audience Research: Use the Thread

A Slack bot for audience research lets your team describe an audience in plain English inside Slack, see how big it is and what defines it, tighten it, and get the export file back in the same thread. Exact Match's Slack bot does exactly that: mention @ExactMatch, ask for something like "solar buyers in Phoenix," and it builds the audience, sizes it, and uploads the export as a Slack attachment, either as a CSV or as hashed files formatted for Meta Ads, Google Ads, or Reddit. Your teammates don't need Claude accounts, and once the bot is enabled anyone in the workspace can use it.

Everyone knows what a Slack bot is. Almost nobody uses the thread properly.

Most people picture a research bot as a search box that lives in Slack. Ask a question, get an answer, move on.

That picture misses the part that actually changes results. With Exact Match's bot, one Slack thread is one working session. Every message you add to the thread builds on the audience you already have. Say "refine it to only homeowners" and it narrows the current audience instead of starting over. Ask "what industries are they in?" and it answers about that same group.

So the thread isn't a chat log. It's the audience, being shaped one message at a time. If a research bot has felt shallow to you before, it probably wasn't your prompts; it was the one-question-at-a-time habit. Teams that treat each request as a fresh question end up rebuilding the same audience five times. The ones that stay in the thread get to the tenth version of an audience instead of stopping at the second.

There are limits to that memory, and they matter (more on those below). But the core habit is simple: one audience, one thread.

What the bot can do

According to Exact Match's Slack integration docs, a single conversation can cover four kinds of work:

• Build an audience from a written description, with geography (cities, states, regions, zip codes, metros), demographics like income, age, and household type, behaviors and interests, and job titles and industries for B2B audiences.
• Start from a list by uploading a CSV of emails, phone numbers, or names, which the bot resolves to people and enriches with traits.
• Analyze an audience: its top signals, income distribution, and household composition, how it compares with another segment, and similar audiences worth a look.
• Export to CSV, or to platform-ready files for Meta Ads, Google Ads, or Reddit with emails and phone numbers hashed for you.

Under the hood it works from the same data as the rest of the platform: 250M+ verified U.S. consumer profiles and 80,000+ targeting clusters. Outside Slack, the same kind of segment building happens in Audience ID.

A workflow that uses the thread well

Here's the order I'd follow for almost any research request. It's cheap in the right places and careful in the right places.

Size it first. Searching for signals, sizing an audience, and previewing it don't draw on your plan. Exports and enrichment do. So start with something like "@ExactMatch How many people are interested in woodworking and live in Colorado?" and get a count before you commit to anything.

Ask what defines it. "What traits define this group?" tells you whether the audience is the one you meant or one that's quietly skewed toward something you didn't ask for.

Narrow it. If the count is too big, ask the bot to suggest ways to narrow it down. It analyzes the current audience and recommends filters.

Export last, and close to launch. This is the nuance most people miss. Interest phrasing like "interested in EV charging" finds people recently researching a topic, and the docs say that signal comes from recent browsing and turns over quickly. Build the audience when you're planning; export it when you're about to run the campaign.

That last step is my strongest opinion here. An intent audience exported two weeks before launch is a worse audience than the same one exported the morning of. The bot makes re-exporting a one-line request, so there's no reason to export early.

The limits worth knowing before you roll it out

Here's where a little-understood tool turns into a frustrating one. These are the bot's documented limits:

Two of these change how you'd run it day to day.

The 12-minute timeout means a thread you walk away from during a meeting is done. Start a new thread and describe what you had ("In the previous thread, I had a 500K audience of solar buyers. Export that to CSV.") and the bot rebuilds it from your description. It can't read the old thread directly.

The 3-conversation cap is the honest limitation for bigger teams. If five people at an agency fire off requests at the same moment, two of them wait. For most teams that's a short queue rather than a blocker, but it's worth knowing before you announce the bot to forty people on a Monday morning.

Getting it into your workspace

Setup is a one-time enablement for your workspace rather than a self-serve install. You request access from Exact Match, receive an installation link, and a Slack workspace admin installs the app. After that there's no per-user licensing; any member of the workspace can use it.

The bot is muted by default. It only responds when someone mentions @ExactMatch in a channel it's been invited to, or in a direct message. Its permissions cover reading messages where it's mentioned, posting replies in the thread, and uploading files. The docs walk through each step, so I won't repeat them here.

One practical suggestion: run audience research in a shared channel instead of DMs. The thread then doubles as a record of how an audience was built, which is useful when a client asks why a segment looks the way it does.

Where the exports land

Finished exports arrive as Slack attachments in the thread, so there's no separate download link to chase. Audiences built from trait filters also show up in your exports list in the Exact Match web app, where you can re-download them or re-export with a different column template without re-running the audience. Some other exports, like location-only audiences, aren't saved in the web app yet, so keep the Slack attachment for those.

Unless you name the columns you want, CSV exports use your organization's default export template, which you set in the web app.

On cost, the bot follows your plan. Exact Match's Unlimited plan has unlimited credits with no monthly cap, so the bot won't quote credits or ask you to approve a spend. It still confirms before a large export, because those take time to run.

If your team also wants agents or scripts pulling audience data outside Slack, the AI agent audience data API post covers that side. For the targeting decisions that come before any of this, see our guide to choosing an audience targeting platform, and if contract terms are part of your evaluation, audience data without annual contract covers them.

And one rule that applies no matter which interface you use: under Exact Match's Acceptable Use Policy, housing, employment, credit, and insurance targeting are restricted uses, and compliance for those categories stays with you as the customer.