Data Tables
Data Tables hold structured records and raw synced provider data your AI can filter, compare, and look up during a run. They're best for structured reference — product and pricing records, lookup tables, and campaign data synced from your ad platforms.
Data tables power conversational agents — they filter, compare, and look up records to answer questions and show product cards — and can also give Studio AI brand context, if enabled. Optimizer agents analyze your synced campaign data directly; they don't draw on manual or reference tables.
What's in a data table
Each table is a set of rows from one source. On the Data Tables page, every table shows its name, an item count, when it was added and last synced, a short description, and the source it's bound to (for example, Meta Ads). Tables are tagged by how they're maintained:
- Raw Sync — pulled automatically from a connected platform (for example, a campaign, ad set, or breakdown table from Meta Ads). These refresh when the source changes.
- Manual — structured records your team added.
- Policy — reference tables used as rules or guardrails.
A synced table may be marked Unmapped until its columns are matched to a standard shape — it still works; mapping just makes it easier to reuse.
Find a table
Use Search, change the sort (for example, newest updated first), and filter by All · Policy · Raw Sync · Manual. The list is paged.
Add a data table
Select New Data Table, then choose how to fill it:
| Source | What it's for |
|---|---|
| CSV upload | Upload a spreadsheet and map its columns — good for product or pricing lists. |
| Manual records | Add rows by hand — best for small, rarely-changing data. |
| API feed | Point at an endpoint Zimmer refreshes on a schedule (Manual, Hourly, Daily, or Weekly) — good for inventory or pricing systems. |
| Categories | Group records into categories to make filtering and lookups easier. |
For ad platforms (Meta, Google Ads, TikTok, DV360), campaign data syncs in automatically when you connect the integration — you don't create those tables by hand.
Columns
A table's columns are its shape. Each column has a key (its identifier, used to map a CSV or API field), a label (the display name), and a type — Text, Number, URL, Image, or Tags — and can be marked required. When you upload a CSV, you map its headers onto these columns.
Link it to an agent
Open an agent's Brain tab and, in the Data section, select the data tables it can use. A conversational agent can use table data to answer questions and show product cards.
Product cards in a journey
In a conversational journey, a Product Card step pulls items from a linked data table.
You set a filter to choose which items — a fixed rule like category = "shoes", or one driven by the
visitor's answers (for example {{age_group}}) — plus how many to show and the display style (carousel, grid,
spotlight, and so on).
Product information that used to live in a separate "Product Catalogs" area is now managed here as a data table.