A 10.7% click-through rate and zero sales: why clicks lied

What this step is

Judging on cost per purchase, not clicks or impressions, is step 49 (Part VIII), marked easy difficulty, often skipped. Easy, because the fix is reading one number instead of three. Skipped, because the wrong number arrives first and feels like progress.

The receipt

15,368 impressions, 1,649 clicks, a 10.73% click-through rate against a normal rate roughly a quarter of that, A$564.35 spent, zero purchases, zero enquiries. The advertisement had promised something the page didn't deliver, so people clicked, arrived, and left. The trap works precisely because the click rate arrives in hours and the purchase arrives in days; there is always a reason to act on the flattering number while the real one is still in transit.

A second receipt, in the other direction

Broad targeting beat a hand-built interest audience on cost per purchase, roughly two to one, across ten ad sets and ten purchases. The raw purchase count favoured broad nine to one, which was the more tempting number to write down, and the wrong one: broad also carried more than four times the spend behind it. Counting purchases instead of cost per purchase is exactly the error this step exists to stop, in either direction.

What done looks like

Every decision read from the business's own payment ledger, cost divided by settled purchases, never from the platform's purchase count and never from click-through rate alone.

Neither receipt above is a one-off anecdote. Both come from the same account's ordinary weekly reporting, which is also why the fix isn't a smarter dashboard, it's reading the one number (cost per purchase from the ledger) that both stories converge on.

Where it breaks

Treating a high click rate as validation when the purchase number hasn't arrived yet. Comparing raw purchase counts between two audiences or creatives without normalising for the spend behind each one.