Amazon Seller & Vendor Central to Power BI

Bring Amazon sales into Power BI with totals you can trace back to the selected report. Start by naming the programme: Seller Central, Vendor Central or Amazon Advertising. Then choose the authorised source, preserve its reporting basis and reconcile the model before sharing it.

The downloadable synthetic Vendor Sales example includes source-shaped JSON, a runnable Python normaliser, a reporting CSV, DAX definitions and expected results. The local checks validate the transformation and arithmetic. No private Amazon extraction, DAX execution or Power BI Desktop/Service test is claimed.

Choose the Amazon source before the connector

Amazon's Analytics Reports reference defines the Vendor report options. The Amazon Ads API is a separate reporting source. A connector supporting Sponsored Ads does not establish that it supports DSP or Vendor Retail Analytics.

Microsoft's Power Query connector directory lists Amazon services such as Athena. These serve a different source from Seller Central or Vendor Central. Check the exact reports, account access and Power BI Service refresh supported by any proposed connector.

Choose a connection route

• A managed connector: useful when its report coverage, history and refresh meet the agreed requirements. Check export access and how changes to previous reporting periods are loaded.
• An owned SP-API extraction: useful when you need control over account selection, stored history and reconciliation. Your extraction handles authorisation, report jobs, errors and maintenance, then stages data in a supported source.
• The teaching files below: useful for checking the model before arranging live account access. They demonstrate a bounded transformation, rather than providing an Amazon connection.

A connector or API supplies data; a warehouse or other staged source stores it. They can form parts of the same design. Our data integration approach covers the reporting architecture.

For a production report job, request the report, check its processing status, then retrieve the completed document and load its verified content. Record the request options and identifiers. A request accepted by Amazon is not a completed extract. Keep access tokens and temporary document URLs outside shared reports and logs.

Define the Vendor Sales example

The example uses GET_VENDOR_SALES_REPORT with these options:

{
  "reportPeriod": "DAY",
  "distributorView": "MANUFACTURING",
  "sellingProgram": "RETAIL"
}

The selected reporting date is 14 September 2026. Amazon specifies Pacific Standard Time (PST) for Vendor Retail Analytics irrespective of marketplace location. Preserve that source date basis when comparing the report with website, order or settlement dates.

MANUFACTURING covers manufactured ASINs regardless of who sourced them to Amazon. It differs from SOURCING. Amazon's Vendor Sales schema identifies ordered fields as populated for the manufacturing view. Keep that option in every reporting key rather than treating the two views as interchangeable.

The download assumes one fictional vendor account, a fictional US marketplace identifier and USD amounts. Its SYN- identifiers are teaching labels, not valid API identifiers. The JSON follows the selected source structure with reportSpecification, salesAggregate and salesByAsin; it is not an Amazon-returned document. The specification's update date and the staging timestamp are illustrative.

Download and normalise the source

Download the complete example ZIP, or inspect the source JSON, normaliser, reporting CSV, DAX measures, expected results and README. Access is ungated.

Extract the ZIP into a local folder. With Python 3 available, run:

python3 normalise-vendor-sales.py

The normaliser reads the supplied teaching JSON and produces vendor-sales.csv. It does not connect to Amazon. Read its source and the README before running it or adapting the mapping to your own extract.

The reporting fact contains one row per vendor, marketplace, ASIN, reporting period and report-option set. The mapping selects salesByAsin, flattens the monetary amount/currency fields and retains source metadata. salesAggregate remains a separate reconciliation control. Appending it to the product rows would repeat the same activity.

The CSV also retains the source's shipped-COGS amount for traceability. It does not establish the supplier's own product cost or a complete profit calculation. The source schema describes revenue amounts as adjusted for returns. Use their reported definitions rather than subtracting returns again, mixing units with money or inferring a cohort return rate.

Import the reporting table into Power BI

1. Choose Get data → Text/CSV, select vendor-sales.csv and open Transform Data.
2. Name the query AmazonVendorSales. Set start_date and end_date to Date, the sample unit counts to Whole Number and monetary amount columns to Fixed Decimal Number. Keep synthetic IDs, currency and report options as Text. Use the README's column list for the complete import mapping.
3. Choose Close & Apply. Add the six definitions below as separate measures.
4. Create cards for the totals and a table by ASIN. Keep the report basis visible beside the figures.

Ordered Units = SUM ( AmazonVendorSales[ordered_units] )

Ordered Revenue = SUM ( AmazonVendorSales[ordered_revenue] )

Shipped Units = SUM ( AmazonVendorSales[shipped_units] )

Shipped Revenue = SUM ( AmazonVendorSales[shipped_revenue] )

Customer Return Units = SUM ( AmazonVendorSales[customer_returns] )

Revenue per Shipped Unit = DIVIDE ( [Shipped Revenue], [Shipped Units] )

Format the unit measures as whole numbers and the revenue measures as USD with two decimals. The sample contains one currency. A live model needs a documented conversion policy before combining currencies.

Check the source-to-report totals

For the stated account, marketplace, date and options, the product rows and source aggregate should both show 25 ordered units, USD 600 ordered revenue, 20 shipped units, USD 480 shipped revenue and 3 customer return units. Check each component; a matching grand total can conceal offsetting errors.

The combined shipped-unit ratio is:

Revenue per shipped unit = USD 480 ÷ 20 = USD 24.00

Product A's ratio is USD 20 and product B's is USD 30. Averaging those gives USD 25, which gives the products equal weight. Dividing the summed revenue by summed units gives the correct combined ratio for this teaching model.

Run the normaliser again. The same input should produce the same two rows and totals. For a live staging load, replace the current rows at the agreed grain, record the source report/update identifiers and retain earlier snapshots separately. Appending repeated reports, overlapping DAY/WEEK totals or both distributor views can duplicate activity.

This exercise checks the selected Vendor retail report. It does not reconcile supplier invoice receipts, Seller Central settlement cash, advertising attribution or profit. Reconcile each additional source separately before modelling the business relationship between them.

Publish and check refresh

For a live build, confirm the account's authorisation, required reports and available history first. Retrieve completed reports into the staged source before refreshing Power BI. Treat failed or unavailable reports differently from verified zero-activity periods, and allow for source updates.

Configure the published connection, credentials and any required gateway. Check Power BI Service refresh history, the source timestamp, row counts and a fixed report total. Microsoft's refresh documentation explains the dependencies; our gateway guide covers the setup checks. A successful local CSV import does not verify a production schedule.

Use shared date, product and account dimensions when adding sources. Keep Seller transactions, Vendor retail metrics and ad reporting in facts with their own grain. Advertising-attributed sales can overlap marketplace sales, so adding the two is not a revenue measure.

For a business selling through its own store as well as Amazon, the Shopify connection guide and ecommerce dashboard guide explain the wider reporting model.

Discuss your Amazon reporting project

Tell us about your Amazon reporting: Seller Central, Vendor Central or both; marketplaces and currencies; the reports you need to compare; and the refresh schedule. Include advertising as a separate source if it is in scope. The link opens an editable integration enquiry.

For delivery scope, see our published BI pricing, US delivery and Australia delivery. If you already have a report to assess, a Trusted Numbers Review examines the source definitions and reconciliation evidence.

Technical sources checked on 17 October 2026. Confirm current access, report availability and options for your account when scoping a live connection.