It correctly produced the order-pipeline result set across the multi-turn flow, including the 93-order stage breakdown and the April-versus-May comparison tables, without manual SQL writing.
What was measured
SQL Generation
Does it generate database-backed SQL correctly?
transformation
What was given, what came back
Test input: Order pipeline breakdown with paid-pending edge case and last-month comparison · text · group: ecommerce-nl2sql-benchmark
Input — what we sent
The exact prompt
How many orders do we have at each stage right now? Follow-up 1: What percentage of our orders were successfully delivered vs cancelled? Follow-up 2: Are there any orders that are pending but already paid? Follow-up 3: Compare that to last month — same breakdown, I want to see if things have improved or got worse.
A deeper operational analysis of current order stages, delivery-versus-cancellation rates, pending-but-paid edge cases, and a month-over-month comparison of the same breakdown.
Why this input is hard
- · Order pipeline analysis
- · Percentage calculation
- · Edge-case detection
- · Payment/order status joins
- · Multi-turn context retention
- · Month-over-month comparison
- · Ambiguity handling for 'same breakdown'
Output — unretouched
Also checked on this input — same tool, 4 other criteria
Business Insight✓ WorkedIt explained what the numbers meant, including the 14% cancellation rate, the completed-orders caveat, the 2 pending-but-paid orders that need attention, and the warning that May is still early because 17 of 21 orders remain pending.Chart / Visualization Support◐ MixedCharts were not generated automatically in the chat; the report says an extra prompt was needed before Definite created a separate dashboard/doc, and the dashboard view shows an 'Orders By Stage' summary with a donut chart.Follow-Up Context✓ WorkedIt remembered the earlier breakdown when asked for 'the same breakdown' last month, reusing the stage categories and the pending-but-paid edge case across turns.Result Readability✓ WorkedThe order pipeline outputs were presented in compact tables with totals and percentages, making the 93-order breakdown and the April-vs-May comparison straightforward to read.
Provenance
- Observation
- 3597b066-7467-4ee7-94c5-45e5c4184ff0
- Evidence run
- db2bb5d5-0e0e-4cb3-8d76-3555c45c23cd
- Study
- Query Live Databases Using Plain English with AI
- Research task
- 86b9y6c99
- Tested at
- not recorded
- Source
- first-party
- Evidence state
- verified
- Proof shown
- input + output shown
- Cost / latency
- not captured
- Repeat run
- not captured
- Tester
- not captured
The last three rows are honest blanks, not placeholders — our capture has no field for them yet.
Query this
get_evidence({
tool: "definite",
scenario: "ecommerce-nl2sql-benchmark"
})MCP · mcp.aidemos.com/api/mcp
Free with attribution.
Same input, same check — 2 other tools
measured on SQL Generation
AskYourDatabase✓ WorkedHandled the month-over-month follow-up with multiple SQL statements to compare the current snapshot, inspect paid-but-pending orders, and build the prior-month comparison.Basedash⚠ StruggledThe first comparison query in the order-pipeline flow hit a small SQL error and had to be fixed and rerun before the answer was produced.
This evidence is published in
From the same study (page rebuilt from a later run)
Real inputs and real outputs, no retouching · every cell queryable via API & MCP · aidemos.com