What is CPA?

CPA (cost per acquisition, sometimes cost per action) is ad spend divided by the number of conversions. If you spent a thousand dollars and got forty purchases, your CPA is twenty-five dollars per purchase. It answers the bluntest question in media buying: what does one customer, lead, or sale actually cost me. Note that "acquisition" is whatever event you optimize for, so a purchase CPA, a lead CPA, and an add-to-cart CPA are different numbers that should never be compared to each other.

The number only becomes useful once you know your ceiling. Your max CPA is roughly your average order value multiplied by your gross margin: the profit a sale generates before ad costs. Pay more than that to acquire a purchase and the sale loses money. Everything in day-to-day optimization flows from that line. Ads below max CPA are profitable and candidates for more budget, ads above it are burning cash regardless of how good their other metrics look. Businesses with real repeat purchase behavior sometimes set the ceiling off expected customer lifetime value instead of first-order profit, which allows a higher CPA today, but that is a deliberate, financed decision, not a default assumption.

CPA and ROAS are two views of the same economics, and buyers tend to prefer CPA when order values are consistent (one flagship product at one price) and ROAS when carts vary a lot, since a fixed CPA target is misleading when one order is worth five times another. Watching both catches problems either one alone would miss, like an ad with acceptable CPA that only ever attracts the smallest carts.

The classic mistakes: judging CPA on a handful of conversions, where one lucky or unlucky sale swings the number wildly; panicking during the learning phase, when CPA is normally volatile; expecting the CPA from a small test to hold perfectly at scale, when some drift upward is normal as delivery expands; and never computing max CPA at all, which reduces optimization to vibes. Calculate your ceiling with the ROAS calculator, then find the ads beating it in Analytics and feed them budget while cutting the ones that never get close.