# What is Nova Data Share?

Source: https://novadata.io/docs/data-share/overview
Checked against Nova's code on 2026-10-29

Nova Data Share is a read-only login to a ClickHouse database, `nova_share`, that holds the data Nova computes for your Amazon accounts as plain tables. Every table is filtered to your organization's rows, so you query your own data from a BI tool, a warehouse, a script or an AI agent, and never see another company's.

## Who is it for?

Teams that want Nova's figures as rows: agencies building client reports, brands joining Amazon with their ERP, data teams filling their own warehouse. If you need answers rather than rows, the app or the MCP is shorter.

## What does your login give you?

- **Read only.** You can query and export, never write.
- **Your organization's rows.** Each table is filtered on your organization. Every seller and vendor account in it is included, and an account you connect later appears at the next refresh.
- **Described columns.** `SHOW FULL COLUMNS FROM nova_share.product_metrics` shows each column's description.

**What is included**

| Access | Tables |
|---|---|
| Included | 18 tables: the P&L (`product_metrics`), today's orders (`orders_live`), the cost dictionary, eight advertising tables including DSP, Search Query Performance weekly and monthly, listing health, inventory, inbound shipments, supply chain, and `dataset_freshness` |
| On request, with a Vendor Central account | the six `vendor_*` tables |
| On request | `orders`, `account_costs`, `sku_costs` |

Your Vendor Central listings and campaigns need no request: they are rows of `listing_health` and of the advertising tables, marked by the `channel` column. A table with nothing for you is empty, not missing: `dsp` without a linked DSP advertiser, `supply_chain` until you fill the Supply chain page. See [Tables](/docs/data-share/tables).

## How do you get access?

In the app, open **Integrations, Nova Data Share** and click **Request access**. Every member of the workspace can see the page. Nova creates your login and sends it to you; Data Share is part of a custom plan. See [Connect](/docs/data-share/connect) for the next steps.

## Is it the same data as the app and the MCP?

Yes. The tables are projections of the data the app and the MCP read. Three differences matter when you compare totals:

- **Raw P&L basis.** `product_metrics` does not apply your presentation choices: spreading monthly storage fees, excluded fees and your own costs stay in the app and the MCP. Its `net_profit` equals the app's net profit when your account uses the default settings. See [How does Nova build your P&L?](/docs/profit-and-costs/pnl)
- **Account costs.** Costs Amazon bills to the account rather than to a product, such as the subscription, are on no `product_metrics` row. `account_costs` lists them line by line, on request.
- **Currency.** Money is in your organization's primary currency, with `marketplace_currency` and the day's `exchange_rate` on the row. See [Currencies](/docs/your-data/currencies).

## How fresh is it?

Nova transforms your data every hour, and each table is copied into the share minutes after its part of the transform, usually between :20 and :55 UTC. Each copy swaps in whole, so a query never sees a half-refreshed table. Three tables keep their own rhythm:

| Table | Copied |
|---|---|
| `orders_live` | every minute |
| `dataset_freshness` | every 2 minutes |
| `orders` | nightly, after 04:00 UTC |

How often Nova reads each Amazon report is in [Refresh times](/docs/your-data/refresh-times).

## How do you know when a table changed?

Read `dataset_freshness`: one row per seller account and table.

| Column | Meaning |
|---|---|
| `dataset` | The table the row describes |
| `cadence` | How often its content changes: `every_minute`, `hourly`, `every_2_hours`, `every_8_hours`, `daily`, `weekly`, `monthly` or `on_change` |
| `updated_through` | UTC end of the latest Amazon window read for the account and present in the share |
| `updated_at` | UTC minute the table last changed in your share for that account |

Poll it every few minutes and pull a table when its `updated_at` is later than at your last pull. Nova's recomputations without a new Amazon read (your cost inputs, exchange rates) do not move it. An account with no Amazon read of a table's reports in the last 45 days has no row for that table.

## Ask your AI

> Using my Nova Data Share, list which tables changed in the last hour for each of my accounts, and how far each one's data goes.