"Enough about average gift. You're obsessed with average gift." -Jeff Goldblum, Independence Day.

Ok, maybe the quote isn't exact. Although if the character were working with non-profits, maybe he would have said this. Average gift is endemic to the charitable sector, and it is reiterated at every corner. It's likely that your fundraising software reports it. It's likely to be in books, blogs, and anything that even remotely references fundraising metrics.

So why do I hate it so much?

With any metric, the first question we should be looking for is, "what are we actually trying to measure?" Average gift is often a measure purported to show how well you are doing as a fundraiser, but that's not really a good quantitative answer. I have seen many reasons to measure average, but the two that seem to come back the most are, "it measures typical giving amounts," and, "it shows how well your major giving is doing."

Think about those two. They are at odds with each other.

There are two factors that cause issue for Average in fundraising as a measure of typical giving. The first is skewness. Without getting too deep in the weeds, you should know that fundraising data tends to be positive skewed. This means that a larger number of donations sit at the smaller amounts, and the volume trickles downward as the gift amounts increase.

That should be fairly obvious. If you were to look at your gifts, I'd bet that 80% of them sit well below $500, but you might have gifts that then cross thousands of dollars, tens of thousands of dollars, or more. That is a positive skew.

A property of positive skewness is that the average ends up not being really centred around your most common gifts. Between the three measures of central tendency, that is, mean (average), median, and mode, average is the highest of the values.

The second problem is centred around outliers. These are defined as non-typical giving amounts, and can be measured a variety of different ways. These outliers bring the average gift even higher, and the shift can be significant, mainly because large gifts are several orders of magnitude larger than typical giving. A single $10,000 gift is 100 times larger than a $100 gift.

So if average doesn't give a good "typical" amount in a number of cases, does it give a good measure of major giving then? The higher the average gift, the more per gift you are getting in general, so it might be tempting to use it for that purpose. However, it can be deceptive as well. For example, let's say you had 10,000 donors of $100 each. You would increase your average gift by around $10 in many scenarios, including:

1) A single new donor giving a $100,000 gift
2) Every donor simply giving an extra $10
3) Losing 1,000 donors of $100 each, but gaining a single donor of $100,000 (in this case, your total giving stays the same while average moves)

Ratios, like average, lack context. Only by knowing total giving and number of donors could we deduce which one of these three scenarios has occurred. This brings up two important points. The first, never report numbers in a vacuum; You should always be reporting on many metrics. The second is then a question you should ask; Am I using the right metric(s) to measure what I want to measure?

Median presents a good option for a typical giving amount. Median is a measure where 50% of your gifts are below that level, and 50% are over. Remember above about the measures of central tendency? This one is the second highest of the three. (I would not recommend using mode for this purpose, because mode will present you with your most common amount, but that can bounce around. Median will be far more consistent when using actual data).

There is also trimmed average (trimmed mean or truncated mean). When you trim, you are removing some gift amounts to average. For example, a trim5 is removing the top, and bottom, 5% of gifts.

As an illustration, I have a dataset where the total amount is around $850,000. The average gift is $425, and the median is $50. If I remove the top and bottom five percent of gifts, and take the average, my new average is $65. If I remove the top and bottom 10%, we get $57. If I remove the top and bottom 20%, we get $49. Notice how we move closer to $50? Median is really a trimmed average where we have removed the top and bottom 50%.

When it comes to measuring major gifts, you can look at total giving to really give you that focus. Unless your number of donors has changed dramatically, total giving will give you a far better idea of how, in general, major gifts has gone. Obviously we can get much more specific with metrics, such as measuring values for only major giving, but it will give you a better idea than average will.

As a note, with the particular dataset, average is akin to the 95th percentile of giving. That is, 95% of gift amounts are below the average amount of $425. That doesn't seem very... average.

Ok, not so fast. Average has its uses, and reporting it is not a bad idea. For example, you can derive insights when you compare changes in averages versus changes in median, total giving, number of donors, etc..However, ask yourself what you are using it for, and if it is the best metric for that purpose.

Disagree with me? Awesome. Let me know in the comments - the best thing we can do is discuss and have a more wholesome understanding of charity metrics. Any questions, anything you would like clarity on, any other metrics you want me to discuss? Let me know! I'm more than happy to talk about more metrics.