A campaign can be briefed correctly, approved correctly and bought incorrectly. The mismatch we saw most often is between an awareness brief and a click-optimised buy, and it hides inside reporting that looks healthy.
The brief said one thing and the buying said another
The brief was about reaching people who did not yet know the category existed. The measure of success was recognition and consideration, over a period of months.
The buy was set to optimise for link clicks, because that is the default in most interfaces and because clicks are what the weekly report contained.
Nobody made an incorrect decision at any individual step. The strategist wrote the brief, the buyer configured a campaign, and the report showed activity.
The two halves were never checked against each other, because there is rarely a point in the process where anyone is asked to do that.
The result was several months of spend that could not have produced the outcome it was funded to produce.
Why the objective setting decides everything
Modern buying platforms are optimisation engines. The objective is not a label on the campaign, it is the instruction that determines who sees the ad.
Given a click objective, the system finds people with a high propensity to click. It does this well, and it will find them regardless of whether they are the intended audience.
Propensity to click is a behaviour, not an interest. Some people click a great deal, across categories, for reasons that have nothing to do with buying.
The system does not know the brief. It optimises for the number it was given, and it will reach that number by the cheapest available route.
Every other setting, including the audience definition, is a constraint on that search rather than a substitute for it.
Clicks are cheap to buy and expensive to interpret
Click volume is the easiest metric in digital advertising to move. Almost any campaign can produce more clicks by relaxing quality.
That makes click cost a poor signal on its own, because a falling cost per click is as likely to mean worse traffic as better buying.
In this account the cost per click fell steadily over the campaign, which was reported as improvement.
Site behaviour told the opposite story. Time on page collapsed and the share of visits that reached a second page fell with it.
Those two signals in combination almost always mean the optimisation has found a cheap audience rather than a relevant one.
The audience the objective selected
Click-heavy audiences skew toward particular placements and particular devices. Low-quality inventory produces clicks in volume, some accidental, some from users who click reflexively.
Accidental clicks are especially common in placements where the ad sits near a control the user meant to press.
An optimiser rewarded for clicks will find that inventory quickly, because it produces the target behaviour at very low cost.
Exclusion lists help and do not solve it. The supply of cheap clicks is large and regenerates faster than any list is maintained.
The only reliable fix is to stop asking for clicks, because the audience follows the objective rather than the targeting.
What awareness buying actually looks like
Buying for awareness means paying for delivery to a defined population at a controlled frequency, not paying for a response.
The relevant controls are reach against the target definition, frequency distribution across the flight, and the quality of the placements the impressions ran in.
Those are less satisfying to report because none of them move quickly, and none of them can be improved by an overnight optimisation.
The measurement is also different in kind. Recognition and consideration are measured by asking people, in a survey structured to compare exposed and unexposed groups.
That measurement costs money and takes weeks, which is why it is often skipped in favour of the numbers the platform supplies for free.
The reporting made the error invisible
The weekly report contained impressions, clicks, click-through rate and cost per click. Every one of those numbers was accurate.
None of them addressed whether anyone had come to know the brand, because the platform cannot measure that and reports what it can.
A report that contains only available metrics will silently redefine the objective as whatever those metrics represent.
Over a few months, the meetings were entirely about click cost. The awareness goal survived in the deck and nowhere else.
This drift is the normal outcome when reporting is assembled from what the tools export rather than from what the campaign was for.
How the mismatch was diagnosed
The tell was the relationship between cost per click and everything downstream. Click cost improved while every on-site measure got worse, in a straight line.
That pattern is diagnostic. Genuine improvement in audience quality moves those measures in the same direction, not opposite directions.
We then looked at the placement breakdown, which had never been reviewed, and found spend concentrated in a small set of low-quality environments.
Finally we checked the objective field, which took a few seconds and explained the whole thing.
Most diagnoses of this kind end at a setting nobody thought to question, because defaults are invisible once they have been running for a while.
What changed when the objective changed
Switching to a reach-based buy made every reported number look worse. Clicks fell sharply and cost per click rose, because the cheap inventory was no longer being bought.
Time on site and repeat visit rates rose over the following weeks, and branded search volume moved for the first time in the flight.
Branded search is a useful proxy for awareness precisely because it is hard to buy by accident. Someone typing the name has to have acquired it somewhere.
The client had to be prepared for the reported decline in advance, because a report that gets worse after a change is difficult to defend retrospectively.
That conversation is the real work in this kind of correction. The setting takes a moment to fix, and the expectation it created takes a quarter.