CPA

CPA is only as trustworthy as the conversion definition and attribution window sitting underneath it, which is why the same campaign can report two different CPAs depending on which report you pull. On Google Ads, CPA is calculated using the conversion action's configured counting method (one per click versus every conversion) and its attribution window, and Smart Bidding strategies like Target CPA optimize toward whatever that conversion action counts, not toward some platform-agnostic notion of a sale. On Meta, CPA (commonly shown as cost per result) is built on the pixel or Conversions API event you chose as the optimization event, using Meta's own default 7-day click / 1-day view attribution window unless you changed it, and since Apple's iOS App Tracking Transparency changes, pixel-only tracking on iOS traffic systematically undercounts real conversions, which pushes reported CPA higher than the true figure without anything in the account actually getting worse.

CPA interacts closely with bid strategy, attribution window, and conversion action setup. Under Target CPA or Target ROAS, the number you see in the report is the actual result of that target, not a separate independent metric - if the reported CPA is running above your target, the algorithm is telling you (through its bidding behavior) that it cannot reliably find enough conversions at that price with current volume, creative, and audience. Widening the attribution window, in either platform, will generally lower reported CPA by crediting more conversions to the same spend, which is a legitimate methodology change but not a real performance improvement, and it is worth knowing whenever a CPA suddenly improves and nobody changed the campaign itself.

CPA matters most once a campaign has real, if modest, conversion volume - enough that a handful of conversions landing or not landing on a given day does not swing the average wildly. It matters least, and is actively misleading, in a campaign's first days of learning phase, where the algorithm is still exploring and CPA can spike or crater based on a small sample, and on any campaign with fewer than roughly 15-30 conversions a month, where the number is closer to noise than signal.

The most common mistake is comparing CPA across Google Ads and Meta as though both platforms measured the same event the same way. A $40 CPA on Google Search and a $40 CPA on Meta Feed are not directly comparable because the underlying attribution windows, conversion counting rules, and the quality of intent behind a search click versus a scroll-stopping feed ad differ substantially. The second common mistake is chasing a target CPA down aggressively on a thin-data campaign, which typically produces volume collapse - the algorithm restricts spend to only the cheapest-looking conversions it can find rather than actually hitting the lower cost at scale.

In reporting, always check which conversion action is driving the number before reacting to a CPA change. A CPA report is only as meaningful as its conversion definition: a campaign counting both add-to-cart and purchase as conversions, or counting every conversion rather than one per click, will show a systematically lower CPA than an identically performing campaign counting purchases only, and neither number is wrong, they are just answering different questions. When CPA moves, check the conversion action, the attribution window, and total conversion volume before assuming the campaign's actual efficiency changed.