Saturday, June 22, 2013

What Your Nonprofit's Twitter Usage May Reveal About Its Fundraising

When it comes to nonprofit fundraising, does the old adage that silence is golden ring true?  Consider the following mini-experiment:
  1. Take the top 100 charities as ranked by Forbes.
  2. Gather data on their fundraising efficiency from Charity Navigator (this excludes some observations due to missing data).  The fundraising efficiency number reflects how much it costs the organization, on average, to raise $1 of donations.
  3. Gather data on how talkative the charities are, as based on their number of tweets.
  4. Consider if/how the fundraising efficiency of organizations correlates with their talkativeness.
To see the data most clearly, I split the observations into quartiles based on number of tweets, from least talkative to most talkative.  The following chart reports the average fundraising efficiency of each quartile.

Note that the least talkative charities spend, on average, less than 5 cents to raise $1; in contrast, the most talkative charities spend 12 cents to raise $1, quite a stark difference.  Not only that, but the cost gradually increases as you move through the intermediate quartiles as well.  In short, organizations who tweet more actively are, on average, less efficient at fundraising.

Statistical Significance
The first question one must ask is whether this is mere coincidence that arises from a small sample.  Using a regression model, with fundraising efficiency as the dependent variable and number of tweets as the independent variable, reveals a positive relationship with a p-value well below 1%.  This means that if there was truly no relationship between the variables, the probability of seeing such a positive relationship in a sample of this size is less than 1%.  That is, the relationship is clearly statistically significant.

  • Is this connection due not to talking (tweeting) but more broadly an organization's overall presence on twitter?  After all, as discussed in this blog before, organizations with more followers also turn out to be less efficient fundraisers.  To address this, I used tweets per follower (rather than tweets) as the "talkative" variable, and the positive correlation persists.
  • Is this connection driven by how an organization splits its costs between fundraising and non-fundraising categories?  After all, many complain that different application of accounting rules is what makes some organizations appear more efficient than others.  To address this, I used revenue growth (rather than fundraising efficiency) as the "fundraising" variable, and the connection again persists.
As with any data exercise, it is important to stress that the statistics demonstrate correlation, not causation.  That is, there is no indication that tweeting will itself reduce fundraising efforts.  However, it does indicate that organizations who actively tweet are also organizations that tend to struggle with fundraising efficiently.  Thus, twitter usage may portend difficulties in fundraising, even if it is not a cause of it.


  1. That is a fascinating analysis. My preconceived guess is the relationship would be the opposite direction.

    As you mentioned, we must constantly keep in mind that correlation is not causation. The breakout raises more questions than it answers. After a few minutes thought I cannot guess at a causal link.

    After looking at the Forbes 100 list again, it would be fascinating to split out those NPOs that have a heavy reliance on GIK. Wonder what the graphs would look like.

    Also wonder if there would be a big split between the larger and smaller of the top 100, say at about the $400M point or so.

    Probably not enough data points to run the analysis by sector (health care, relief & development, GIK, social services).

    1. Jim,

      Excellent points. Following your suggestion, I repeated the exercise excluding GIK charities (food banks and the like) as well as with only the top 50. In each case, the correlations were very similar. Would be interesting to view at the sector level with a larger data set.

      As for why this connection is there, I can only guess. My instinct is that it may have something to do with the overall outreach approach of the charities. Twitter is impersonal by nature so may be further down the pecking order of ways to reach donors. Those reliant on twitter, then, may have a more spread out donor base. Would be curious to hear what you and others think may be driving this.

  2. Thanks for running the numbers again.

    Seems like there would be some underlying factor explaining what the gaphs point out. I’m drawing a blank on ideas.

  3. Possible positive interpretation: The high-Tweeting orgs are spending more on investment and innovation; this leads to lower CPDR, but has potential in future payoff. Possible negative interpretation: The high-Tweeting orgs have trouble focusing on the basics, are spending resources on "shiny objects" like Twitter etc.

    These two interpretations may not be mutually exclusive.