Skip to main content
Use metrics to locate the stage that needs attention. Visits describe traffic arriving at tracking. Clicks describe tracked offer-click activity. Conversions describe the events your campaign marks as conversions. The difference between those stages is why two conversion rates can both be correct and still show different percentages. All formulas below use the same filtered report scope. If you compare two metrics from different dates or campaigns, the ratio no longer describes one funnel.

​
Read one funnel from arrival to profit

A campaign has 1,000 Visits, 200 Clicks, 20 Conversions, USD 300 Revenue, and USD 100 Cost. If CTR falls while CR is stable, investigate the journey before the offer click. If CTR is stable while CR falls, investigate the offer and conversion stages. First confirm that both comparisons use the same tracking configuration and sufficiently comparable traffic. To reproduce the calculation, select one campaign and period, add the five base columns, then add CTR, CV, CR, ROI, and Profit Margin. Check the base counts before questioning a derived percentage.

​
Counts and monetary totals

Conversions is an event count. If two stages for one order both count as conversions, the order contributes two. Suspicious % describes the tracker’s visit flag, not rejected orders or a guaranteed measure of fraud.

​
Percentage formulas

The API returns fractional ratios for these built-in rates. For example, API 0.2 is displayed as 20%. Repeated conversion events can make a conversion rate exceed 100%.

​
Earnings and cost per action

For the example campaign, EPV is USD 0.30, EPC is USD 1.50, and CPA is USD 5.00. The report’s CPC is USD 0.50 because it divides by 200 tracked offer clicks. A source that bills 1,000 ad clicks uses a different denominator.

​
Event-specific columns

Each Event ID from 1 to 10 has four columns: For Event 2, use event2, revenue2, cost2, and value2. These columns let you compare stages without marking all of them as conversions. Indexed amounts retain recorded values even when inclusion flags or cost corrections make overall totals different.

​
Interpret unique metrics carefully

Metricanic builds visitor identity from the campaign, normalized IP address, and normalized user agent. Distinct counts use approximate aggregation within the selected scope. This is useful for separating repeat activity from distinct campaign visitor identities. It does not establish the number of individual people. A changed IP or browser can create another identity, and the same person in another campaign belongs to another campaign identity. Historical records without usable identity data can count as events without adding to the unique total. Do not add unique counts across overlapping groups. The same identity can appear in several dates or offers while counting once in the combined scope.

​
Distinguish zero from an undefined rate

If a denominator is zero, the built-in derived calculation returns null in the reports API. The interface currently formats missing or non-finite metric values as zero. A displayed 0% ROI with zero cost therefore does not establish that the campaign broke even. Totals recompute rates from total counts and amounts. They do not average row percentages. One row with 1 conversion from 2 visits and another with 9 from 90 visits produce total CV 10 / 92 × 100 = 10.87%. When a percentage looks wrong, check its denominator, event configuration, active filters, and whether the table rounded a small amount. For your own formula, see Custom metrics.