Which tables are in Nova Data Share, and how do you read them?

Nova Data Share holds 27 tables: 18 included for every organization and nine on request. Each answers one question at one grain, and a few rules on days, currencies and cuts keep your sums right.

View as Markdown Ask Claude Ask ChatGPT

Checked against Nova's code on October 29, 2026

All docs

What am I earning?

Table	One row is	Use
`product_metrics`	account × marketplace × ASIN × SKU × day	P&L lines, units, traffic, BSR, reviews
`orders_live`	order line, current UTC day and the three before	What sold today; `is_provisional = 1` until the report confirms it
`orders` (on request)	order line, last 24 months	Each sale with fees, refunds, settlements
`account_costs` (on request)	settlement line not tied to an order line	Subscription, storage lumps, inbound, removals, coupons
`sku_costs` (on request)	settlement line on a SKU, without an order	Reimbursements, clawbacks, reversals
`cost_dictionary`	Amazon cost key, the same for everyone	Block and category of each cost

Where is my ad budget going?

Table	One row is	Use
`advertising`	ad structure × attributed ASIN × day	Console view of Sponsored Products, Brands, Display
`advertising_campaigns`	campaign × day (× ad group × ASIN on Products, Display)	Campaigns, longest history
`advertising_targets`	target × day, Brands and Display	Audiences, keywords and product targets
`advertising_placements`	campaign × placement × day	Top of search, product pages, rest of search
`advertising_search_terms`	ad structure × search term × day	What shoppers typed
`advertising_negative_targets`	negative keyword or target, current state	What each account blocks
`advertising_impression_share`	keyword × snapshot day, covering 30 days	Impression share and rank, as in Campaign Manager
`dsp`	day × line item × `cut`	Amazon DSP display and video

What is the market searching for?

Table	One row is	Use
`sqp_weekly`	marketplace × parent ASIN × search query × week	Market demand and your share
`sqp_monthly`	the same, by month	Same, longer history

What is the state of my listings and stock?

Table	One row is	Use
`listing_health`	account × marketplace × SKU, current state	Listing status, issues, Buy Box
`inventory`	account × stock region × SKU × day	FBA and AWD stock
`inbound_shipments`	FBA shipment × SKU, current state	Sent and received
`supply_chain`	stock region × ASIN, current state	Your own counts, supplier orders, lead times

What about Vendor Central?

On request, for organizations with a Vendor Central account.

Table	One row is	Use
`vendor_product_metrics`	vendor account × marketplace × ASIN × day	The 1P P&L, on `product_metrics` names
`vendor_inventory`	vendor account × marketplace × ASIN × week	Amazon's stock of your products
`vendor_purchase_orders`	purchase order line	What Amazon ordered and received
`vendor_invoices`	invoice	Invoices and their settlement
`vendor_remittances`	payment line	What Amazon paid and deducted
`vendor_forecast`	vendor account × marketplace × ASIN × forecast week	Amazon's demand forecast in units

dataset_freshness (account × table) says when each table last changed: see Overview.

Which days and currencies do rows use?

Sales use the marketplace's local day, advertising the Ads profile's time zone, and _utc columns UTC: see Connect.

Money is in your organization's primary currency (currency). marketplace_currency is what Amazon transacted in, and exchange_rate the rate of the row's day. exchange_rate does not add up: never sum() it. To return to the marketplace currency, divide row by row, then aggregate: sum(product_sales / exchange_rate).

Unconverted: advertising bids and budgets, Search Query Performance prices, and Vendor Central purchase orders, invoices and remittances (each in its document's currency).

Which tables can you add together?

Never sum across cuts. advertising_campaigns, advertising_targets, advertising_placements and advertising_search_terms cut the same spend as advertising another way. Take each total from one table, never two.
DSP is separate. Never add dsp to advertising. Inside dsp, add rows within one cut only, and never on the audience cut. product_metrics.dsp_spend is already part of dsp.
Vendor and seller P&L. Aggregate vendor_product_metrics and product_metrics separately, then add the P&L columns.
Levels are not flows. active_subscriptions, star_rating and the weekly stock of vendor_inventory are states: read one value, never sum over days.

How long can figures change?

Tables	Window that can still change
`product_metrics`, `orders`	about 60 days (returns, fees, settlements)
Advertising tables	about 14 days; the current day is excluded
`dsp`	about 14 days; the closed month is read again on the 5th and the 20th
`inventory`	the current day only
Current-state tables	replaced at each refresh
`advertising_impression_share`	never: each snapshot is fixed

Ask your AI

From my Nova Data Share, compare last month's Sponsored Products spend in advertising with ppc_spend in product_metrics, by marketplace, and explain any gap.

Something missing or unclear? Write to contact@novadata.io. New to Nova? Start a free trial.