September 28 to October 4, 2026
$600Total order amount- Completed orders
- 4
- Average order amount
- $150
Restaurant data, put to work
Did sales grow because you had more orders or bigger orders?
A weekly total tells you whether sales went up. Separating order count from average order amount tells you how. Here is a small example you can check yourself before connecting your restaurant.
Synthetic example. These are not customer results.
One more order. The same average spend.
One example location, one POS, two complete Monday-to-Sunday weeks. Dates use New York time. Amounts are US dollars.
October 5 to October 11, 2026
$750Total order amount- Completed orders
- 5
- Average order amount
- $150
25% higher total. 25% more orders. No increase in average order amount.
The restaurant recorded one more completed order. Average order amount stayed at $150, so the higher total came from order count, not a higher average order amount. These figures do not establish why the extra order happened.
The next question is where that additional business came from. This comparison does not tell you whether an ad, a repeat guest or something else caused it. That needs more evidence.
The math behind the answer
Previous average order$600 ÷ 4 orders = $150 per order
Current average order$750 ÷ 5 orders = $150 per order
Change in total order amount($750 − $600) ÷ $600 × 100 = 25%
When combining reports, divide the combined amount by the combined order count. Do not average the reports' averages when they contain different numbers of orders.
What exactly do these numbers include?
Sum of interface.order.amount, in dollars, for completed Square order rows at one location. The Square mapping subtracts tax, service charge and tip from the order total. This is not profit, an accounting net-sales definition or marketing-attributed revenue.
COUNT of non-null id values on the selected completed Square order rows. This single-source example has one row per order. Check row identity before reusing the method for another POS.
For a real restaurant, confirm the import is complete and reconcile the same location, dates and definition with the POS. Missing records are not evidence of zero sales.
Try the example with your AI.
Download the sample CSV, upload it to your AI and use this request. The sample includes records outside the selected location, source, status and dates, so the answer should apply those filters rather than add every row.
Download the synthetic sample CSVCompare September 28 to October 4 with October 5 to October 11, 2026, in America/New_York. Use only COMPLETED Square rows for synthetic-square-location. Show total amount in USD, completed orders, average order amount and the change between weeks. Exclude other records. Explain whether the difference comes from order count, average order amount or both. Do not infer why it happened.
The expected totals are $600 and $750. A matching result checks this example, not a live connection to your restaurant.
Inspect the exact query recipe
The public walkthrough includes the Feast data tools, source definitions, date boundaries and expected results. Its queries are checked against the actual query endpoint with synthetic records.
Now ask about your restaurant.
Feast's free connector lets a supported AI client work with your available POS data, without repeating the export for every question. Setup confirms your POS, available history and AI client. Start with one useful comparison, then build from there.
Start my free setup