How metrics work. CollectionsIQ stores only additive base components — impressions, clicks, spend, conversions, conversion value, units sold, gross sales, discounts, refunds, orders. Every ratio (CPM, CTR, CPC, ROAS, CVR, AOV, net sales…) is computed from those sums at evaluation time. This matters: averaging pre-computed ratios gives wrong answers; summing components and dividing once gives correct ones, for any window and any scope.
Two data families, kept honest.
- Ad metrics come from what Google and Meta report — including their attributed conversion value. Per-platform ROAS is that platform's reported value divided by its spend; blended ROAS sums across platforms first.
- Shopify sales metrics come from your real orders. Net sales = gross − discounts − refunds. This is your ground truth and is never silently mixed into ad ROAS.
- Combined metrics bridge the two deliberately — e.g. true ROAS (Shopify net sales ÷ ad spend) or cost per unit (spend ÷ units sold).
Custom metrics. On the Metrics screen you can define your own formulas over any base components — margins, blended efficiency scores, whatever your business runs on. A custom metric behaves exactly like a built-in: usable in rules, previews, and thresholds. Formulas are validated when you save, so a broken formula can't corrupt a rule.
Guards. Any condition can carry minimum-data guards (e.g. "at least 500 impressions", "at least $50 spend") so a product with 3 clicks and a lucky sale doesn't outrank your real winners.