PDQ's analytics help you make data-informed decisions about checkout performance and A/B tests. This is your glossary.
OEC (Overall Evaluation Criterion) - the single metric used to judge a test. Traditional metrics conflict (raising AOV via a higher free-shipping threshold can lower CVR; discounts raise CVR but lower AOV), so PDQ uses ARPC as the unifying OEC.
Primary metric - ARPC (Average Revenue Per Checkout): Total Revenue ÷ Checkout Sessions, which equals (AOV + ASR) × CVR. When cost data is available: Gross Profit per Checkout (Revenue − COGS) ÷ Checkouts and Direct Profit per Checkout (Revenue − COGS − Shipping Cost) ÷ Checkouts.
Core formulas
Metric | Formula |
CVR | Orders ÷ Checkouts |
AOV | Revenue ÷ Orders |
ASR | Shipping Revenue ÷ Orders |
ARPC | Total Revenue ÷ Checkouts = (AOV + ASR) × CVR |
Gross Profit / Checkout | (Revenue − COGS) ÷ Checkouts |
Direct Profit / Checkout | (Revenue − COGS − Shipping Cost) ÷ Checkouts |
Component metrics - Checkout extensions report Impressions (distinct checkout tokens/day who saw it), added items, and upsell revenue.
Track360 metrics - revenue from tracking-page visits, visits & avg visits/order, orders stuck in transit (>7 days), and order failures.
Significance testing
ARPC → t-test for difference in means (compute SE from each group's SD, then t, then p; p < 0.05 = significant).
Conversion rate → two-proportion z-test (pool CR, compute SE and z; p < 0.05 = significant).
SRM (Sample Ratio Mismatch) → chi-squared; χ² > 3.841 (95%, 1 df) means the split is off and the test may be invalid.
Takeaways: use ARPC as your primary metric; validate with t-tests (ARPC) and z-tests (CR); watch for SRM. Need help interpreting a test? Reach out to your CSM.