Two emails a month. A call every six weeks. One congress touch per cycle. Those numbers usually come from a planning document written by someone who has since moved on, and they have rarely been revisited against response data.
Method, stated up front
Geometric carryover fitted per channel across anonymised specialty brands, with a Hill response shape and ridge regularisation. Recent periods held back on a rolling origin. Where two channels flighted together closely enough that their coefficients could not be separated, they are reported jointly.
What separates
Fit carryover per channel and the half-lives pull apart immediately. Personal promotion frequently carries for several weeks. Approved email often spends out inside one. Peer programmes carry longest and tend to be the most under-flighted relative to how long their effect persists.
A test worth running
Take the channel that carries longest and widen its interval in one matched region.
Put it into the channel that spends out fastest, in the same region.
Read both at twelve weeks. Earlier reads are mostly noise and can talk you out of a correct decision.
One region gives a direction. A point estimate implies more precision than a single region supports.
Worth stating: these are fitted values across a sample, not constants. Your own carryover depends on customer mix, competitive position and how consistently the channel was actually executed. Use these as a starting prior and fit your own.