Is cost per conversion the same as cost per acquisition?

Google defines cost per conversion (“Cost/conv.”) as how much, on average, each of your conversions cost, “calculated by dividing your total cost by the number in your ‘Conversions’ column”. It adds that the calculation only applies to eligible interactions, so clicks that cannot be tracked for conversions are removed from it. The Target CPA bid strategy is built on the same quantity: when you select it, Google says, “you set your desired average cost per conversion”. Google's own documentation expands the letters as cost per action. Whichever expansion you prefer, in Google Ads the number you are reading is that one division.

Where the two really part company is the denominator. The Conversions column counts conversions across your primary conversion actions, and Google notes it may include modelled conversions where not every conversion can be observed. Which actions count is a setting: by default most conversion actions are included unless someone unticks them, a campaign can be given its own conversion actions, and the counting setting records either every conversion after an interaction or only one. So an account counting enquiries, calls and purchases in one column reports a cost per conversion below its cost per customer acquired, and both figures are correct about different things. Before comparing a cost per conversion with anyone's cost per acquisition, ask what each one divides by.

Google dates both halves of its division, the spend and the conversions, by the click. It reports the primary conversion columns against the time of the click, not the time of the conversion, so a conversion this week from a click last week is reported in last week, which Google says is what keeps cost per conversion aligned with spend. The consequence for anyone reading a recent window is that its conversions can still grow against clicks already paid for, so a cost per conversion for the last few days can fall after you have read it.

What Riibon does with it. Cost per conversion equals cost per click divided by conversion rate, exactly, provided all three are computed from the same spend, clicks and conversions and conversion rate is taken over clicks. Riibon computes them that way for that reason, rather than reading Google's Conv. rate column, which Google calculates over eligible interactions such as ad clicks or video ad views. And we do not report the figure on its own in a client table, because a movement in it cannot tell you whether the price of a click moved or the rate at which clicks convert moved, and those take different fixes, usually owned by different people.

When a cost per conversion moves, Riibon splits the movement between those two halves, using each factor's own log ratio against its own baseline. If one half carries 65 per cent or more of the movement it is named as the driver; otherwise both are. If the two halves' movements together total under about 5 per cent, it refuses to attribute at all, because a which-half answer that small is noise dressed as a finding. A worked illustration, not a client result: a click price up 10 per cent and a conversion rate down 40 per cent give a cost per conversion up about 83 per cent, and the split puts about 84 per cent of that movement, measured on a log scale, on the rate. A bid change there works on the smaller half.

One comparison goes wrong often enough to name. Google says to judge a Target CPA strategy against its average target CPA, the traffic-weighted average the strategy actually optimised for, rather than the target you typed in, because ad group targets, device bid adjustments and changes to the target over time all move it. Compare an achieved CPA against the typed target and a strategy can look off target when it was aiming somewhere else.