Learn & Understand

Why the Same Campaign Can Report Three Different CPAs

CPA math itself - cost divided by conversions - is straightforward. What's not straightforward is that "conversions" isn't a single, objective count; it depends entirely on which attribution model is counting them, and different models can produce meaningfully different CPA figures from the exact same underlying customer data.

Last-Click Attribution: The Simplest, and Most Limited, Model

Last-click attribution gives 100% of the credit for a conversion to whichever ad or channel the customer interacted with immediately before converting, ignoring every earlier touchpoint entirely. This is the simplest model to implement and understand, but it systematically undercounts the contribution of upper-funnel channels - a display or social ad that first introduced a customer to a brand gets zero credit under last-click if that same customer later converts after a final search ad click, even though the earlier ad may have been essential to the eventual purchase.

Multi-Touch Attribution: Spreading Credit Across the Journey

Multi-touch attribution models instead distribute conversion credit across several touchpoints in a customer's journey, using various weighting approaches - linear (equal credit to every touchpoint), time-decay (more credit to touchpoints closer to conversion), or position-based (extra credit to the first and last touchpoints specifically). Under a multi-touch model, an upper-funnel awareness channel typically shows a meaningfully higher share of credited conversions - and therefore a lower calculated CPA - than the same channel would show under strict last-click attribution, even though the actual campaign spend and real customer behavior haven't changed at all.

View-Through Attribution: Crediting an Ad That Was Never Clicked

Beyond click-based attribution entirely, view-through attribution credits a conversion to an ad the customer simply saw (an impression), without ever clicking, if that customer converts within a defined attribution window afterward. This is common in display and video advertising, where most exposure never results in a click at all - view-through attribution captures influence that a click-only attribution model would miss completely, but it also introduces more judgment about how much causal credit a mere impression genuinely deserves.

Why the Attribution Window Length Also Matters

Every attribution model also requires choosing a time window - how many days after exposure a conversion still counts. A 30-day window will credit meaningfully more conversions (and therefore produce a lower CPA) to a given channel than a 1-day window, simply by counting a longer stretch of post-exposure customer behavior as attributable, independent of any change in the ads themselves.

Applying This When Comparing CPA Across Channels or Reports

Before concluding one channel is more cost-efficient than another based on CPA, confirm both figures were calculated using the same attribution model and window - comparing a last-click CPA for one channel against a multi-touch or view-through-inclusive CPA for another is comparing two different definitions of "a conversion," not a fair apples-to-apples efficiency comparison.