I think there are two observations that we can make using this diagram. First, three of the four months with the largest amounts of money spent were in the year before the November 2016 election. As might have been expected, October 2016 had the highest total cost ($8,588), but the other three months with the largest expense—June, September, and December 2015—were all before either of the presidential primaries took place. This makes sense, though, when we consider that the first month the IRA started creating these ads was, indeed, June 2015. So the IRA spent larger amounts of money in its incipient stages in order to establish the foundation for its campaign, creating Facebook groups and building up followers. Once these various groups had accrued thousands of members, it would cost less to reach the same numbers of people since they were already subscribed to these groups. In June 2015, the IRA paid an average of about $85 per ad and generated an average of 113 impressions per dollar; in October 2016, they paid an average of about $33 per ad and generated an average of 458 impressions per dollar; in April 2017, they paid an average of about $9 per ad and generated an average of 1,267 impressions per dollar.

The second observation gleaned from this tree map, and the one that may stand out the most from this visualization, is that the month with the highest number of ads created by the IRA is not in the election year of 2016, as one might assume; rather, April 2017 featured the highest number of ads created by the IRA. At first blush, this might not seem noteworthy. As shown above, by this point the IRA had well-established Facebook groups that were reaching significant numbers of Americans, but why the surge in ads during this month? The previous month had similar metrics, approximately $20 per ad and $1,006 impressions per dollar spent, but with almost half as many ads. So I looked at major news stories from April 2017 to see if anything groundbreaking happened. While the swearing in of Neil Gorsuch as an Associate Justice of the Supreme Court (April 10) and the federal judge striking down Trump’s executive order to pull federal funding from sanctuary cities (April 24) were noteworthy events that may have led to an uptick in Russian ads—particularly the latter given how many ads had already fomented xenophobic sentiments—there was one other event that I think may have had an effect. On March 30, Michael Flynn, former National Security Advisor to Trump who had resigned in February for providing misleading information to the FBI regarding his meeting with the Russian ambassador, offered to testify before Congress in exchange for immunity. My conjecture is that IRA saw this news and decided to try to employ some misdirection to divert attention away from a story that could potentially lead back to them.

Of course, this is purely conjectural, but I think the rapid decline in IRA Facebook ads after May 2017 lends some credence to this hypothesis. There were 0 ads created in June 2017, 3 in July 2017, and then 1 in August 2017. Something made the IRA pull the plug on this campaign rather quickly. Most likely, it was the May 9 dismissal of James Comey from his position as Director of the FBI and the May 17 appointment of Robert Mueller as the Special Counsel to oversee the investigation into possible Russian interference in the 2016 election. I believe that these two observations, taken together, demonstrate that the IRA was closely watching the U.S. election cycle and current events to time their ads at key moments, and ultimately to jettison the campaign when the investigation began.