Growth.

The framework

1. 01

   The decay curve you already have the data for.

   You don't need a study, you need one chart per channel. Take the primary response metric — clickthrough, reply rate, cost per acquisition, conversion from that source — and plot it by month since the day you started running it. What you're reading is the slope, and roughly where it crosses the level at which the channel stops paying for itself. Most operators have never drawn this, because each month's number gets looked at against last month's, and that is precisely the resolution at which a steady decline is invisible.

2. 02

   Is this fatigue, or is it the law?

   When a channel softens, the reflex is to blame the creative and rebuild it — and sometimes that's right, because a worn-out ad genuinely does recover with a new one. The distinction is whether the floor moved. Creative fatigue shows up as your numbers falling while the channel's underlying costs hold steady. Structural decay shows up as the costs themselves rising: auction prices up across the category, competitors describing the same slide, the platform changing what it rewards without telling anyone. Refresh the creative once. If the second refresh doesn't hold, stop arguing with the channel.

3. 03

   Judge a young channel on its slope, not on today's number.

   Two mornings ago the instruction was to switch the others off and put everything behind the winner. It stands — concentration is a spend decision, not a knowledge decision, and a successor test is the cheapest insurance you will ever buy. But judge it properly. A new channel almost always looks worse than a mature one on the day you compare them, because the mature channel carries years of accumulated optimising and the new one carries three weeks. Compare them that way and you kill every successor before it grows. Give it a defined run, then read the direction of its numbers and where they could plausibly land.

4. 04

   Your org chart is quietly a bet on one channel.

   Depreciation is not only a media problem. Over a few years the team, the tooling, the agency relationships and the hiring criteria all specialise into whatever has been working, so by the time the channel is visibly finished, the organisation built around it can't move. That is why the switch costs a year rather than a quarter. Keep one person genuinely fluent in something you barely spend on, and treat that fluency as insurance rather than as slack you could reclaim next time budgets tighten — the successor test above needs somebody who already knows how to run it.