Return on Ad Spend

Deciding whether to keep an ad running means putting what went out and what came back on the same line. The numbers easiest to count on the advertising side are impressions and clicks, and those aren't denominated in money.

So the meeting misfires. The person running ads reports 1.2 million impressions and 9,000 clicks, and the CEO asks whether any of it makes money. The first set of numbers can't answer that, because several steps sit between "clicks went up" and "revenue went up."

ROAS skips those steps and reports a multiple. Put in 1 of spend, get out this much revenue — which means the conversation about raising or cutting budget starts in the same unit.

Divide the revenue from an ad by what was spent on it. Spend $2,000 and produce $10,000 and it's 5. Some places write that as 500 percent; same number.

Three things need settling first.

1. Which revenue counts as belonging to that ad. This is the number of days after seeing the ad within which a payment still counts, and seven days versus thirty is a large difference.
2. How far the revenue is counted. First payment only, or subsequent renewals too. On a product billed monthly, counting only the first month pushes the value far below reality — 1.2 on a first-month basis frequently becomes 3.4 across a year.
3. Revenue or profit. This measures revenue, so cost of goods and salaries aren't in it. A 5 with a 70 percent cost ratio leaves almost nothing.

Calculate it per channel and don't average them together. Three channels summing to 3 doesn't tell anyone which one to expand. Set it beside customer acquisition cost (CAC), what it takes to win one customer, and lifetime value (LTV), what that customer leaves behind, and you can see whether the multiple is one that holds.

"ROAS is 5 and our cost ratio is 70 percent, so there's nothing left."

The revenue multiple alone doesn't say whether anyone made money.

"It read 1.2 counting first payments. With renewals it's 3.4."

What got counted as revenue flips the conclusion.

"Break it out by channel. The total doesn't tell us what to pull."

A combined number doesn't determine the next action.

At Pagedeck, Ravi put channel-level ROAS side by side in the quarterly report. One channel read 4.8 and another 1.1, and the two combined came to 2.9. The previous quarter, looking only at the combined figure, nobody caught the weak one.

Grace asked for cost of goods and the six-month renewal rate in the same table. The same revenue multiple leaves different amounts behind, and customers who pay once are mixed in with customers who stay a year.