How do I read out A/B test results with iDBQuery?

An A/B test readout compares two variants to see which performed better and whether the difference is real rather than noise. iDBQuery lets you analyse your own experiment data directly instead of exporting it to a stats tool.

• You'd ask iDBQuery: 'For the checkout experiment, compare conversion rate, sample size, and average order value between variant A and B, and show the lift with a confidence read.'
• It splits the data by variant, computes each metric and the difference, and returns a comparison chart plus a table.
• Every figure is cited to its source rows, so 'variant B converted at 4.8%' is backed by the actual assignments and outcomes.

iDBQuery keeps the readout honest by showing its SQL and the sample sizes, so you can judge whether a result is solid or underpowered, and it will note when a difference is small relative to the noise. Follow-ups flow in one thread: 'Does B still win for mobile users?' or 'Check for a novelty effect by week.' Because experiment assignment, conversions, and revenue may live in different systems, the one live model joins them with no pipeline. Save the readout as a shareable report for the product review, and let the Analyst agent probe whether the win holds across key segments, citing the rows. This complements funnel, correlation, and segmentation analysis.