Picking up a sample size of customers the statistics way?

Dear members,

I have a large group of our customers who have transacted with us once.

I would like pick up a sample group of customers from this available group mentioned above for an experiment.

The experiment I would like to do is I would like to gain insights in their behavior as to transact with our organization again over the next 1 year or so.

Through our CRM systems I can easily find out every single month whether or not a past customer transacts with us again or not. Note that I have to do the exercise manually for each customer even though the CRM will help me on this.

The guidance I need is out of the large group of our customers who have transacted with us once, how do I arrive at a sample size so that the sample represents the large group of customers which are available with me, since it is not feasible for me to do manually do the exercise of repeat purchase behavior using our CRM for all of them.

Your guidance will definitely help.


TS Contributor
How often are transactions made per year per customer in general? How large of a group can you "afford"? (tests in marketing often costs money since they often come out with a negative result. I mean, the company already uses near best practice and if a practice that is going to be tested is so good, why isn't implemented it already?) Are you only interested in customers who has only made one and only one transaction? What exactly is it that you want to find out?

Thank you for the reply.

I would like to find out the pattern of repeat purchase for those customers who have transacted with us at-least once in the past.

As to find out whether or not a customer who has transacted with us in the past does a repeat transaction with us, I have to use our CRM, where in I have to MANUALLY KEY IN EVERY TIME the customer details as to find out whether he has made a transaction with us month after month.

Since it is virtually impossible for me to do the above exercise for the HUGE list of customers it is logical for me to think of a SAMPLE which will represent the population which is
why I have raised the query.

Please let me know if there is a better way (or) how does one arrive at the SAMPLE.


TS Contributor
If you're doing it manually then it's a question of time. I don't understand a few things though, and which you don't have to answer. Aren't frequency of transactions implemented within your CRM model? If not, why? Frequency of transactions is usually a major constituent of a CRM model as far I know. The other thing I don't really understand is why you have to do it manually for each customer. Once the code is written it can usually be used for all items. Which software are you using?

We use a CRM module which has been fully created in house.

Yes, I do acknowledge that a CRM module should give answers to queries itself readily, but because the CRM system (panel may be the right word to use instead of system) which is made in house, it is not that sophisticated one, which is why I have to use the panel manually as to finding out on the transaction record.

Because I cannot do the manual check for the total database available I wanted to pick up a sample which may represent the total data base, so that I arrive at the patterns of their repeat transactions.

Because I wanted to arrive at a sample, I had raised the query here at this forum.