The tutorials and most examples in this documentation use Retail Chain Operations, a fictional retail chain with 20 stores in five Chinese regions, 54 products and about 31,700 order lines from January 2025 to August 2026. Load it once and every example can be repeated on your own server.

You need permission to upload file datasets and analysis models (an administrator account works).

Download the files

File	Contents
retail-chain-operations.xlsx (3.7 MB)	One sheet per table: dim_date, dim_store, dim_product, dim_customer, dim_channel, dim_employee, fact_sales_line, fact_store_month, fact_service_ticket, plus a _README sheet with a field dictionary and check totals.
retail-chain-operations-model.zip	The Retail Chain Operations analysis model (tables, relationships, hierarchies and 23 measures such as Net Sales and Gross Margin Rate), bound to the data source retail_chain_operations.

Amounts are in CNY. All names are generated; the data describes no real company or person.

1. Upload the workbook as a file dataset

1. Open Data › Datasource and select the File datasets tab.
2. Under Upload a file dataset, click + on the CSV / Excel Files card and choose retail-chain-operations.xlsx.
3. In the upload dialog:
   • Dataset: enter retail_chain_operations exactly. The model refers to its tables by this name.
   • Sheets: select every sheet except _README.
   • Table: keep each table name identical to its sheet name (dim_date, fact_sales_line, …).
4. Click Upload and wait until the dataset's Status turns green. fact_sales_line has the most rows and finishes last.

See File Dataset for the dialog fields.

2. Upload the model

1. Open Models and click Upload.
2. Choose retail-chain-operations-model.zip with Please select upload file, then click Save.

The model appears in the list as Retail Chain Operations and uses the data source retail_chain_operations, which is why the dataset in step 1 must have exactly that name: the model also refers to the tables as retail_chain_operations.<table>. If you used another name, delete that dataset and upload the workbook again with the right name. A model with the same ID that already exists is replaced; see Managing Analysis Models.

3. Check the numbers

Build a quick report on the model, for example a Measure card or a Table, without filters, and compare these totals over all dates (the _README sheet lists more):

Measure	Expected total
Net Sales	5,094,084.39
Sales	5,797,364.69
Order Count	13,596
Paid Order Count	12,388

If a total differs, the most common cause is a sheet that was skipped or uploaded into a table with another name: check the dataset's table list.

Optional: register the metrics

The model contains measures but no Metrics Library bindings. To try governed metrics and the AI Agent's metric answers, open the model in the modeler, open the Metric bindings tab at the bottom and click Generate metrics from this model; see Metrics Library.

Next

Create Your First Analysis Report builds a chart on this model.