# Forecasting Membership Lives

#### DarcyDarcy

##### New member
Hi All,

I am currently attempting to forecast how long a pool of members will stay with a golf club for on average.

I have suggested simply taking the average number of years which each current member has been with the club, a friend has suggested taking the average number of years people who have left the club where there for?

Assuming all data is available, what would be the best way to estimate how long the current group of members will stay with the club on average?

Much appreciated

#### lev888

##### Full Member
I would agree with your friend.

#### JeffM

##### Elite Member
Hi All,

I am currently attempting to forecast how long a pool of members will stay with a golf club for on average.

I have suggested simply taking the average number of years which each current member has been with the club, a friend has suggested taking the average number of years people who have left the club where there for?

Assuming all data is available, what would be the best way to estimate how long the current group of members will stay with the club on average?

Much appreciated
This is really not math unless you have a huge body of data. A basic idea in inferential statistics is that, other things being equal, more data are more reliable than fewer.

I'd average the number of years that every new member stayed. If you count only those who have dropped out, you will probably underestimate likely longevity. If you count only those who have not yet dropped out, you will probably overestimate longevity.

If the five current members have been around on average for 11 years, but eight more joined and stayed in only an average of 1 year, 11 seems too high and 1 seems too low.

#### Jomo

##### Elite Member
Hi All,

I am currently attempting to forecast how long a pool of members will stay with a golf club for on average.

I have suggested simply taking the average number of years which each current member has been with the club, a friend has suggested taking the average number of years people who have left the club where there for?

Assuming all data is available, what would be the best way to estimate how long the current group of members will stay with the club on average?

Much appreciated
I agree with lev88. Say the average current member has been there for 8 years then this number is too low since they may on average stay anther 5 years. Then the avg is 13 and 8 is obviously too low.

#### tkhunny

##### Moderator
Staff member
Hi All,

I am currently attempting to forecast how long a pool of members will stay with a golf club for on average.

I have suggested simply taking the average number of years which each current member has been with the club, a friend has suggested taking the average number of years people who have left the club where there for?

Assuming all data is available, what would be the best way to estimate how long the current group of members will stay with the club on average?

Much appreciated
You might be able to do something useful on whatever scale you have. Survival models arewell-studied by actuaries, medical professionals, and many other fields of endeavor.Decrements can be anything from Death to simply failing to continue in the experiment.
Don’t be alarmed by the calculations. This should be easy for a competent spreadsheet user.

1. Take ALL your data. Include anyone who ever was a member.
2. Simply count those who paid 1st yeardues and fees and call it N0
3. Count all those who paid 2nd yeardues (or continued their membership into the 2nd year) and call it N1

Continue this process with all the data you have.
It is likely that you will have to DECIDEwhat the last year is.

4. Once you have N0, N1, N2, …NLast, calculate the annual survival rates.

1. S1 = N1/N0
2. S2 = N2/N1
3. Etc…
4. Slast = NLast/N(Next to Last)

5. Once you have all of these, plot them on a chart and see if it looks anything that might be remotely rational. If you have giant bumps or troughs, you will have to decide if you should leave it that way or smooth it.

1. A trough might be caused, for example, because you once had a special on 5-year memberships when paying in advance. You might expect a trough in S6.
2. A bump might be caused by a special membership drive. Maybe, “Join One More Year at 50% Discount”. This might be harder to spot,since you might have 1st years, 4th years, 10th years, etc buying into the campaign.

6. One more thing… Create another series of values…

1. T1 = S1 = Probability of Surviving to the end of the 1st Year.
2. T2 = S1*S2 = T1*S2 = Probability of Surviving to the end of the 1st Two (2) Years.
3. T3 = S1*S2*S3 = T2*S3 = Probability of Surviving to the end of the 1st Three (3) Years.
4. TLast = T(Next to Last)*Slast = Probability of Being the Longest Tenured Member Ever!

7. Anyway, even with relatively small data, you may produce something that is useful to you. If you can figure out how to input your personal judgment with the empirical data (Step #5), you may be able to improve the model. Of course, if you’re a dreamer, you may make it worse.

8. Finally, your expectation of membership length is simply S1 + S2 + S3 + … + Slast. You may wish to add ½ for theoretical reasons.

Does it look like fun?
I think so.

Last edited: