Wikipedia:WikiProject Wikidemia/Fundraising

From Wikipedia, the free encyclopedia

This page organizes statistical analysis of WMF fundraising efforts, with a focus on trying to understand patterns in contributions that will be useful for designing future fundraising campaigns. Content on this page is derived from the Wikimedia official fundraising information and raw data.

Please place suggestions or requests for figures or numbers that you would find particularly useful to create and compute on this page's talk page.

Inferring lessons from these retrospective data is quite challenging. In order to understand better what fundraising techniques are more or less effective, prospective evaluations in the form of field experiments like these proposals would be invaluable. Such evaluations would complement existing literature.

For the three fundraising drives from 2/17/2005-2/28/2005, 8/18/2005-9/8/2005, and 12/16/2005-1/20/2006, I focus on USD denominated contributions, and I summarize the number of contributions each day, the total dollar value of contributions each day, and the average size of contributions each day.

Comparisons of the 2005 Fundraising Drives[edit]

First, the following three figures compare the time paths of the three fundraising drives, up through 13 days of the most recent fund drive.

USD value of donations by day
Number of donations by day
Average donation size by day


Note that up through this point, the most recent fund drive resembles quite closely the previous two fund drives. Average contribution size is roughly flat over the course of all of the fundraisers, and almost identical from one drive to the next. In all three fund drives, the number of contributions and the total value of contributions ramp up over the first couple days and then erode steadily over time. The next three figures extend the data through 1/20/2006, past the official 1/7/2006 end of the most recent fund drive.

USD value of donations by day
Number of donations by day
Average donation size by day


Now the conclusions are potentially quite different. Though earlier fund drives fizzled over time, the most recent fund drive exploded with success in the days around New Year's. More than USD$10,000 was given on 7 out of the 8 days from 12/30/2005 to 1/6/2006.

In all of these fund drives, the contributions have predominantly been quite small. Only 7 contributions of >$1000 have been made, and only 20 of >$500. Less than 2% of contributions have been for $200 or more:

Histogram of Donation Size, All Fundraisers


2005q4 Fundraising Drive[edit]

Analysis/figures related to the 2005q4 fundraising drive. These figures examine more closely the trends over the course of each week.

Here are graphs of the total amount of USD donations, the number of USD donations, and the average size of USD donations since the start of the drive:

Total dollar value of donations by day
Number of donations by day
Average donation size by day


General Conclusions: The average size of contributions was slightly larger during Week #2 of the fundraising drive than during the (holiday) Week #3. More notable is the steady decline in the daily number of contributions over Weeks #1-#3, from over 400 per day at the start of Week #2 (after a two-day ramp-up) down to an average of just over 100 per day during Week #3. This was followed by the blockbuster Week #4. The holidays confound interpretation of these trends, but it would surely be valuable to figure out what happened to turn things around in Week #4. (Jimbo's personal appeal was made at 19:33 on 31 December 2005.) One question is whether it might be more effective to have frequent, short drives than rare, extended ones; another question is what changed about the type of appeal during Week #4 that led to more contributions. Note that the number of daily contributions since the official end of the fundraiser has been only slightly lower than the number of contributions during the first three weeks of the fundraiser. It would also be useful to know what causes fundraisers to produce sustained generosity of this type.

2005q3 Fundraising Drive[edit]

Analogous procedures and plots for the 2005q3 Fundraising Drive resulted in these figures:

Total dollar value of donations by day
Number of donations by day
Average donation size by day


General Conclusions: The average size of contributions was essentially constant during this fundraising drive. The daily number of contributions peaked at over 350 in the middle of Week #2, fell to a level of about 100 per day, and stayed constant at that level through the end of the drive.

2005q1 Fundraising Drive[edit]

Analogous procedures and plots for the 2005q1 Fundraising Drive resulted in these figures:

Total dollar value of donations by day
Number of donations by day
Average donation size by day


General Conclusions: It's somewhat more difficult to draw conclusions about this fundraising drive because of its relative brevity. However, it appears that the average donation amt was steady in the $25 range throughout the drive, except for a large spike on the final day; and the number of donations per day fell relatively slowly during the drive after a two-day ramp-up at the start.

Procedures[edit]

The figures above were generated using the following procedures (more details available upon request):

  • I cut and pasted all of the automatically generated db content on contributions, eg from

12/18/2006 and 12/19/2006 into a single text file.

  • I deleted every line in that file that didn't represent a contribution. The result for a portion of the 2005q4 fundraiser is this data file. To facilitate future work, it would be convenient if a data file like this could be automatically generated from WMF's donation tracking system.
  • Then I used this analysis file to begin studying the data using the Stata software package.
  • I used Stata to generate and export the graphs.

Analogous figures can easily be made for contributions denominated in other currencies, for contribution totals across all currencies, and for more recent contributions. Roughly 2/3 of contributions, and of the total dollar value of contributions, were in USD.