Learn and master Clay with Clay university

Here's a question almost everyone hits: when do I use an Audience, and when do I use a Table? The confusion is understandable, because the enrich action inside an Audience shows the same familiar table view you already know. But because Audiences operate at a completely different scale, what you use them for is different too.

‍⚡ Update: enrichment now runs on Workflows. When this lesson was recorded, orchestration lived in Tables and a bulk enrich was the bridge to get there. That bridge has moved. Enriching a segment now builds a Clay Workflow, and the branching that used to need a table can happen inside the enrichment itself. The bulk enrichment is still available as Create enrichment table. If you've never built a bulk enrichment, you'll only see the workflow path. See Enriching an Audience with Workflows.

🧭 The Quick Answer

Match the job to the surface. This table is the decision matrix in short form:

‍

The single constraint behind most of this: a Clay Table caps at 50,000 rows. Anything larger, including any "millions of records" segmentation, belongs in an Audience.

🔬 Audiences Are Production, Tables Are the Laboratory

The cleanest way to hold the difference in your head:

• Audiences are for scale and for the source of truth. Segment millions of records, store permanent information (firmographic, demographic, or custom research you already know works) with a bulk enrichment, take simple final actions like sending to a sequencer or Slack, and keep your CRM current through an export sync. Static, reliable fields like domain, location, and lookalike group are perfect to enrich at the Audience level.
• Tables are your testing ground. This is where you iterate, form a hypothesis, and test it, especially for multi-step flows with lots of conditional runs orchestrating actions across columns. You rarely want to do that experimentation at 50,000-row scale.

Experiment in a Table, then graduate to scale in an Audience. Once you've figured out your data structure, you shouldn't need to wire columns together with conditional runs. You should be able to segment at the company or people level and run one enrichment. Done right, moving to scale makes things simpler, not more complex.

🔁 Connecting the Two

Your Audiences and your Tables share one unified data layer, so data flows both ways. Open the tools menu in a Table and search "audience" to find four audience enrichments: you can pull in audience data the same way you pull from a CRM, update audience records, and upsert into an Audience.

Table → Audience (Upsert)

When you've tested something in a Table and you're confident in it, push it back to your Audience.

1. Open tools and choose the upsert into Audiences enrichment, then pick the Companies (or People) audience.
2. Set your lookup fields to match on domain and/or professional social URL, so you don't create duplicates.
3. Map the fields to update, for example name and industry.
4. Save and run a small batch first (10 rows) to confirm it works before running the whole table.

The matched records land in your Audience. Filter the Audience and you'll find the rows you just sent, with their fields populated.

⚠️ One Thing That Trips People Up

• Always test on a small batch first. Run 10 rows before you commit the full audience. The small run is your safety check.

Put it together and the default rules are simple:

• Data you know is secure, stable, and reliable: run it in the Audience.
• An experiment, a hypothesis, a Claygent scrape you're unsure about, or a multi-column conditional flow: test it in a Table first, then deploy it back.

🚀 What's Next

You now know which surface fits which job: Audiences for scale and the source of truth, Tables for experimentation and multi-step orchestration, and the upsert bridge that connects them.

Next, put it into practice with the use-case lessons, where these same moves power CRM enrichment, TAM sourcing, signal-based plays, and AI outbound at full audience scale.