The calculation is total earnings over a period divided by clicks in the same period. Send 2,000 clicks to an offer and earn 900, and your EPC is 0.45. Its power is normalization: a placement sending 500 clicks and one sending 50,000 can be compared on the same axis, and an offer's EPC can be set directly against the cost per click of the traffic feeding it. If a source costs 0.30 per click and returns 0.45, the placement carries a positive margin; flip those numbers and no volume will save it.
EPC is also where the deal model quietly distorts the picture. On CPA, earnings land at the FTD and EPC stabilizes within days. On RevShare, revenue accrues over months, so the EPC of a young campaign systematically understates its true value, and comparing a week-old RevShare link against a CPA link by EPC alone will always favor the CPA. Meaningful comparisons fix the attribution window, or use a projected EPC built on cohort lifetime value rather than cash received so far.
The metric hides composition, too. A strong blended EPC can mask one geo or landing page doing all the work while others lose money, which is why EPC belongs at the segment level, per offer, per source, per geo, per page, not as one site-wide number. Used that way it becomes a routing signal: smartlinks and traffic allocation decisions are, in the end, an exercise in sending each click to its highest expected EPC.
Why it matters
EPC is the bridge between traffic cost and deal revenue, the number that tells you whether a click is worth buying and which offer deserves it. But it is an average built on FTDs and NGR upstream; treat it as the output of those fundamentals, not a target in itself, or you will optimize toward cheap clicks that never deposit.
Related