Database Modelling and Segmentation Remember philosopher Karl Popper’s dictum that ‘If the theory is true, then the prediction is true’. However importantly Popper said you can’t prove a theory is true or false, you can prove however that a prediction is true, and therefore trust the theory as a result. 

This section is going to talk briefly about regression analysis and database modelling. This is to explain where the best segmentation equations come from. They come from mathematical modelling of thousands of databases over many years. They have proven their worth and they will apply to your database.

Importantly, leave regular donors to one side for a moment. We can show you similar models for regular donors that will focus on retention, upgrade and other critical factors for your regular giving program. No matter how important regular donors are to you, lets focus for a moment on the often neglected appeal or mail responsive donors.  A linear regression of the donor data in your database produces a model that can be expressed as:  
If Yi is the likelihood of a donor giving to an appeal request, then Yi is determined by a number of major variables (called dominant variables) and variables of lesser importance, finally followed by factors than cannot be expressed (such as the mood of the donor at the time of receipt of your letter) which are treated as noise. Your database will reveal that the donors most likely to give again are ‘above average in gift’ donors who have given multiple times and who have given recently. This is called the Recency-Frequency and Amount model and is always the best expression or model of your donor’s relationship with your organisation. It is not the only model but for general donors, i.e. non-regular donors, it is the best model.

So in the formula above β0 might be recency, β1 could be amount and β2 could be the number of gifts the donor has given.

You will find that a β3 parameter for the above formula would be interest in the issue. There is a noticeable propensity for donors to give again, where they have given to an appeal on a similar issue. If they have been thanked for the original gift and ideally if they have been informed as to progress made as a result of their previous support. It is important to record the theme or issue of all appeals.

Let NGO SYSTEMS analyse and target your programs using our tools developed over 35 years in a variety of organisations and countries. Remember all our services are supplied on the basis that if you follow our advice you will raise more funds, or we won't charge you.
 

Note: As a simple example, if you mail regularly to a database of between 4-5,000 donors spread over three years of currency (time since last donation) you should raise consistently more than $50,000 per appeal mailing. Mail six times a year and you should generate more than $300,000 in donations even from this small database. If you are not, ask yourself why?

Your costs should never be much more than about $1 per item mailed including postage, leaving you with about $9 per item mailed in net income from each appeal. If you are mailing an expensive and heavily designed pack, you may well be suppressing your appeal response, the key is always under design appeals, in a thoughtful way.