The situation
Paid search stops performing. The team concludes the channel is saturated, winds it down, and moves the budget to a new one. The new channel works for about six weeks, which is roughly how long novelty lasts, and then settles at the same disappointing place the old one did.
This can repeat for a year. Each cycle produces a plausible post mortem about auction dynamics, or algorithm changes, or the channel being wrong for the category.
What the company usually believes
That channels decay and the job is to keep finding new ones. This belief is sometimes correct, which is what makes it durable and expensive.
Which of the three it is
A positioning problem, presenting as an acquisition problem.
How the other two get ruled out
Not a downstream leak. Accounts that arrived through referral and through the founder network converted and retained at a much better rate than accounts from either paid channel. Same product, same onboarding, same everything after the click. What differed was what the person believed on arrival.
Not a lack of urgency. The referral cohort bought promptly. The paid cohort never got far enough to hesitate.
The evidence
The comparison is the whole diagnosis. Hold the message constant and vary the channel, and if the result is the same in both, the channel is not the variable. Compare a cohort that arrives already understanding the offer against one that arrives cold, and the gap tells you what the copy is failing to do.
A channel that underperforms everywhere is not a channel problem. It is the same problem, visible in more than one place.
What we do
We stop treating channel selection as the lever. The claim being carried through the channel gets rewritten first, against the buyer the referrals are already bringing, and only then does the channel get judged again on its own merits.
What changes
The original channel becomes viable again, because what is carried through it changes. Whether it is the best channel is a separate question, and now an answerable one, because the message is no longer the confound.
What the company stops doing
Rotating budget between channels on a six week cycle. Writing post mortems about auction dynamics for a problem that lives in the copy.
The lesson
Channel performance is usually downstream of positioning. If a message is not working, moving it to a new audience does not fix it, it just costs a fresh budget to discover the same thing.