This is a request for comments on a list membership model I'm pulling together. It might be useful for people who are involved at all in email list management and need to put some rough and ready forecasts together. There's a slight chance that it has some use outside that group (in theory, I suppose, it could work for any situation where there are members).
Anyway -- all I want in return is criticism, hints about how you may have approached a similar situation, and a wish list fro what you'd like to see on it.
If you're used to using Excel, it should be simple to read/adapt for your own purposes. If you *DO* use it for your own thing, I'd appreciate it if you'd let me know.
ANYWAY:
Assuming that:
The probability of churn is highest in a given member's first month than in subsequent months; and that the longer a member stays on the list, the less likely they are to unsubscribe.
Unless one mismanages a list very badly indeed, a hard core of each month's recruits will remain with the list. This assumes that the list remains relevant to its members over a long period (a football newsletter, for example, as opposed to a wedding newsletter)
Then:
The rate of churn for a given period of recruits will decay over time, until the list owner is left with a stable group of core users.
For example, assuming a single recruitment period, a list owner might lose half their recruits in month one, a quarter of the remaining recruits in month two, followed by an eighth of the remaining recruits in month three.
The rate at which this happens could be called the "churn decay" rate. In the example above, the churn decay rate is 50%
This rate of churn will be different for each period of recruits/members, but not so different that we can't assume a common initial churn and churn decay rate.
Please download it and have a play if you're into this sort of thing.
MatM
P.S.
One could (of course) flip these rates on their heads, and call them "retention rates". But since the actual consumer action being measured in each case is the rate of unsubscribes, I thought it might be better to keep it this way round.
At a certain level, I'm also counting non-opens and bounces as "churn".
Digital Planning at Porter Novelli
10 November 2003 16:18pm
This is a request for comments on a list membership model I'm pulling together. It might be useful for people who are involved at all in email list management and need to put some rough and ready forecasts together. There's a slight chance that it has some use outside that group (in theory, I suppose, it could work for any situation where there are members).
Anyway -- all I want in return is criticism, hints about how you may have approached a similar situation, and a wish list fro what you'd like to see on it.
If you're used to using Excel, it should be simple to read/adapt for your own purposes. If you *DO* use it for your own thing, I'd appreciate it if you'd let me know.
ANYWAY:
Assuming that:
The probability of churn is highest in a given member's first month than in subsequent months; and that the longer a member stays on the list, the less likely they are to unsubscribe.
Unless one mismanages a list very badly indeed, a hard core of each month's recruits will remain with the list. This assumes that the list remains relevant to its members over a long period (a football newsletter, for example, as opposed to a wedding newsletter)
Then:
The rate of churn for a given period of recruits will decay over time, until the list owner is left with a stable group of core users.
For example, assuming a single recruitment period, a list owner might lose half their recruits in month one, a quarter of the remaining recruits in month two, followed by an eighth of the remaining recruits in month three.
The rate at which this happens could be called the "churn decay" rate. In the example above, the churn decay rate is 50%
This rate of churn will be different for each period of recruits/members, but not so different that we can't assume a common initial churn and churn decay rate.
I've built a simple model of this (and put it up on http://www.mediaczar.com/downloads/modelling_v.1.xls)
Please download it and have a play if you're into this sort of thing.
MatM
P.S.
One could (of course) flip these rates on their heads, and call them "retention rates". But since the actual consumer action being measured in each case is the rate of unsubscribes, I thought it might be better to keep it this way round.
At a certain level, I'm also counting non-opens and bounces as "churn".