Most campaigns that fail were never going to work, because the offer was wrong before any money was spent on reaching people. Targeting and creative get the blame, but they are the later problems.
The offer does most of the work
A campaign is a way of putting an offer in front of people. If the offer is not compelling to the audience you can reach at a price you can afford, no amount of creative execution rescues it. Before spending anything, be able to state plainly who this is for, what changes for them, and why now rather than later.
The “why now” is the part most campaigns skip. Without it people agree the product sounds good and do nothing, which looks like a targeting failure in the data and is not.
Reach the people closest to buying first
- Existing customers. Cheapest to reach, most likely to convert, and routinely ignored in favour of new audiences.
- People who have already shown intent. Visitors, trial users, abandoned carts, newsletter subscribers.
- Lookalikes of your best customers. Built from actual purchase data rather than from everyone who ever visited.
- Cold audiences. The most expensive, and the right place to spend only once the closer layers are exhausted.
Measurement, and the ways it lies
Attribution is where campaigns are misjudged in both directions. Last-click reporting gives all the credit to whatever the buyer touched last, which systematically overvalues search on your own brand name and undervalues everything that created the demand. People who were going to buy anyway get counted as campaign conversions.
The practical correction is not a better attribution model but a holdout: withhold the campaign from a comparable group and compare outcomes. It is the only method that answers the question you actually care about, which is what the campaign caused rather than what it was present for.
What to watch while a campaign runs
Leading indicators tell you whether to keep spending before revenue data arrives. Click-through rate tells you whether the message lands. Landing page conversion tells you whether the promise survives contact with the page. Cost per qualified lead tells you whether the audience is right. If the first two are healthy and the third is not, the problem is usually the offer, not the media.
Common questions
How long before judging a campaign? Long enough to cover your normal purchase cycle. Judging a considered purchase after a week measures noise.
Should I discount? Discounts reliably raise conversion and reliably train customers to wait. Use them to clear stock or to break a specific hesitation, not as a default lever.
How much should I spend testing? Enough to reach a decision, which means enough conversions to distinguish real differences from noise. Underfunded tests produce confident conclusions from nothing.
Why did a campaign that worked last year fail now? Audiences saturate and competitors copy. The offer that felt novel becomes the baseline, and the same creative against the same people delivers less each time.
Join the discussion