The Cost of a Lost Phone Number

Today I did a very fun math project. I calculated the value of finding a new phone number for our Telefund program to call. I was hoping to demonstrate that the dollars it takes to find new phone numbers are more than repaid by the dollars raised by having additional prospects to solicit.

And, they are -- many times over. This project yielded the best investment opportunity I've seen recently!

This seems like a no-brainer, but sometimes you need to show your work to motivate action. And, of course, making decisions based on data is generally a good idea. You tend to learn some surprising stuff, in addition to confirming suspicions. For example, I was surprised to see that the value of a found number varied widely between different areas of my institution.

So, how did I do it?

First, I calculated the net attrition of records with phonable numbers. (How many less numbers do we have this year than last year?)

Then, I projected how much revenue was lost because we weren't able to call those records. I multiplied the number of lost records by the participation rate (percentage of donors who give) of that particular pool of donors, and multiplied that result by the average gift size of that pool.

Gross lost revenue = number of lost records * participation rate * average gift

Net lost revenue = (number of lost records * participation rate * average gift) - cost of finding a new record

Okay, so now I know how much revenue could have been had. Then, I simply divided this lost revenue by the number of lost records, yielding the average cost per lost record. Or, to phrase it positively, the average potential revenue gained by finding a record.

Average value of a lost record = lost revenue/lost records

Then I made a few lovely charts, one of which was a bar graph showing more and more lost revenue each year due to phone number attrition, and including a projection for next year. It literally resembles low-hanging fruit, due to its inverted nature: since it displays negative numbers, the bars actually hang below the x axis.

All in all, this was a fun and easy project, once I figured out how to conceptualize it. I would love to hear from other people who have undertaken similar projects. Comment, please!

(Some caveats -- this was a fairly rudimentary stab at this concept. I didn't control for people who had opted out of being called (as opposed to numbers simply lost), nor did I consider whether people with lost numbers are likely to have different participation rates and average gift sizes than those whose phone numbers we retain. That's a more complicated project for a different day.)