Marketing metrics are only useful when teams agree on what they mean, how they are calculated, and what grain they belong to. This glossary defines common marketing metrics with formulas, aggregation notes, and pitfalls to watch for when building dashboards, exports, or semantic-layer datasets.
Paid media metrics usually arrive from ad platforms at account, campaign, ad set, ad, date, device, or country grains. They become safer when source definitions and currency behavior are explicit.
Amount spent on paid media during a reporting period.
Number of times an ad was served or shown.
Number of unique people exposed to an ad or campaign according to a platform's identity rules.
Average number of impressions per reached person.
Cost per thousand impressions.
Number of ad clicks, link clicks, or tracked click events.
Click-through rate, or the share of impressions that generated clicks.
Cost per click.
Analytics metrics depend on identity, sessionization, consent mode, event definitions, and property settings. They should not be treated as exact matches to ad-platform clicks.
Visits to a site or app within a measurement window.
Unique visitors or users measured by an analytics platform.
Users first seen during a period according to the analytics platform.
Share of sessions with no meaningful engagement under a specific analytics definition.
Share of sessions or users meeting a defined engagement threshold.
Funnel metrics need stable event IDs, qualification rules, and dates. A lead created date, qualified date, and accepted date can all answer different questions.
People or organizations that submitted a qualified contact signal.
Cost per lead.
Marketing-qualified leads that meet defined qualification criteria.
Sales-qualified leads accepted for sales follow-up.
Share of users, sessions, clicks, leads, or another eligible population that converts.
Cost per acquisition or cost per action, depending on the conversion being measured.
Revenue and profit metrics should state whether they are gross, net, attributed, recognized, order-date based, refund-date based, and currency-normalized.
Sales value generated during a period.
Revenue returned per unit of ad spend under a stated attribution model.
Marketing efficiency ratio, often total revenue divided by total marketing spend.
Revenue before discounts, refunds, returns, and some adjustments.
Revenue after selected deductions such as discounts, refunds, or returns.
Revenue remaining after cost of goods sold.
Gross profit as a share of revenue.
Average order value.
Customer metrics depend on identity resolution, first-seen logic, cohort windows, and whether value is revenue-based or profit-based.
Number of completed purchases.
Customers making their first purchase or first qualifying conversion.
Customer acquisition cost.
Share of customers who purchase again within a defined repeat window.
Share of customers retained over a defined period.
Estimated or observed lifetime value of a customer.
Time needed to recover acquisition cost.
Metric Hive treats metric definitions, grain, currency, and aggregation behavior as part of the data contract. These rules are useful even before a team adopts a formal semantic layer.
Keep raw spend, clicks, impressions, orders, revenue, and costs available so ratio metrics can be recomputed at the reporting grain.
Campaign-day metrics, order-line metrics, customer metrics, and cohort metrics should not share a dashboard without clear labels.
Qualification rules, cost rules, attribution windows, and currency policies change. Versioning keeps historical reports explainable.
Most reporting mismatches come from aggregation, identity, attribution, or unstated denominator choices.
Grain is the level at which a row is true, such as campaign-day, ad-day, order, order line, customer, or cohort. Metrics should only be joined or aggregated when their grains are compatible.
Rates such as CTR, CPC, CPA, ROAS, margin, and AOV are weighted by their denominators. Averaging row-level ratios can produce incorrect totals.
ROAS uses attributed revenue divided by ad spend. MER usually uses total revenue divided by total marketing spend. MER is a blended efficiency metric, not a channel attribution metric.
They are unique counts. The same person can appear in multiple days, campaigns, or properties, so summing rows can double-count people.
CPA can refer to any paid action or acquisition event. CAC should refer to the cost to acquire a new customer under a defined acquisition-spend policy.
The safest answer is a governed revenue definition that states whether it is gross, net, order-date based, refund-adjusted, and currency-normalized.
Use the ecommerce profit glossary for contribution margin, order profit, payment fees, shipping cost, CAC, LTV, inventory cost, and product-level profitability terms.