What End User Experience Looked Like During the Presidential Election by Team AppNeta November 7, 2012
Filed under: Case Studies
I think it’s safe to say that everyone knows what happened yesterday – the U.S. Presidential Election. And while we were interested to see who would be named the winner, we were also interested in seeing how the election would affect end user experience of certain popular websites: mittromney.com; barrackobama.com; foxnews.com; nbcnews.com and cnn.com. Our AppView Web solution uses a zero-admin microAppliance to regularly perform HTTP requests between remote sites and important web services. User experience is broken out by server, network and browser response times and is trended over time in the charts below.
We configured AppView Web to monitor these five sites starting on Monday and in to today. However, we noticed the most server latency occurred between 12:00PM EST yesterday until 12:00AM EST this morning. We saw some interesting spikes in server delay that correlated with specific key moments in the election.
Let’s start with the candidates’ sites.
We saw various peaks and valleys throughout the day while monitoring to mittromney.com – specifically in the mid-to-late afternoon. Presumably, this traffic could have come from the undecided voters who were conducting last minute research on the candidates. Here you can see browser (pink), network (green), and server (blue) response times to a client HTTP request over time. The vertical gray bars indicate brief intervals where a full response to our HTTP requests wasn’t received.
There was an uptick in server latency throughout the evening corresponding with poll closing times, starting at 7:30PM EST. The greatest time events were between 9:20PM and 9:30PM EST, when Florida was going back and forth by the narrowest of margins between both candidates, but had swayed to Gov. Romney. Also at the end of this time frame was when Wisconsin had been awarded to President Obama.
At 11:15PM EST leading up to the official announcement, there was a noticeable amount of traffic still causing spikes in the server response. You can see a gradual, but significant drop in response times at 11:30PM EST when the race had been awarded to President Obama.
We saw an overall greater amount of server latency for President Obama’s website. Throughout the afternoon, possibly for the same presumed reasons as we saw on Gov. Romney’s website, we saw a steady amount of utilization, dropping off at 5:00PM EST.
However, when polls started to close at 7:30PM EST, there were a steady upswing, peaking just after 9:00PM EST when Pennsylvania was awarded to President Obama. There are corresponding peaks and valleys as the states were awarded to either candidate, the largest of which came at 11:30PM EST when the race was called by the major news outlets.
Now let’s look at some of the major news sources’ sites.
We noticed the first deluge of server response latency begun at 1:00PM EST and ended abruptly at 5:00pm EST. Throughout the evening as states were being awarded, there were latency spikes similar to those on the candidates’ site at 9:00PM EST when Florida was hanging in the balance and when Pennsylvania was awarded to the President. The pinnacle was again at 11:30PM EST when the race was awarded to President Obama.
Interestingly, nbcnews.com and foxnews.com had the same pattern of heavy afternoon usage. NBCnews did offer online video or mobile devices though their site, which is evident by the amount of response time on the server end throughout the evening hours. There is the same build up to the announcement at 11:30PM EST and then a steep drop after that to what looks to be normal traffic.
Compared to the other news sites, CNN saw varying spikes in server response time throughout the afternoon. There are the noticeable peaks and valleys in traffic throughout the 9:00PM-11:30PM EST timeframe when the major states were awarded, ending with the official announcement.
We found it so interesting to take a look at user experience differences during yesterday’s events on these various sites. It was exciting to match up the peaks and valleys with the real-time events of the day and see how everything is related.
Did you notice any peaks or valleys in your own user productivity yesterday?