The unexpected has happened: One day, India became the 9th country with the most visits in the hispanic site Taringa. This meant that suddenly there were millions of users visiting a Hispanic site from Asia. Surprised, the directives of the site asked me to find an explanation. What answers could be found? What was going on?
Looking into the case
When something like this happens, the first thing is to understand who the users that are behaving like this are and what are the implications.
Thanks to the Google Analytics graphs, we created a advanced user panel of people connecting from India and within seconds we had our first conclusion: The change had been drastic and sudden.
We were still far away from understanding this behaviour, although our initial hypothesis was that there was some sort of bot that would enter the site unrecognized by Google Analytics as such. We were not ruling out some sort of virus or malicious software in the PC’s of users entering the site. One of the most disconcerting facts was that the percentage of blocking was within normal values.
The following discovery made, came from a geographical report:
Almost all the visits came from the same city in India, Pune. Searching through Wikipedia, we knew that it was the seventh largest city in the country with over 4 million and a half inhabitants. According to the tendency and patterns, in less than two months we would have 100 per cent of the inhabitants of Pune visiting out site. Something didn’t make sense. We searched without finding any clues if there was conferences or massive reunions happening in that city in the following months. We couldn’t find anything.
The next step was to analyze the visiting sources. According to confidentiality rules, I am not sharing the graph but the conclusion we got was the following: The distribution between visiting sources (users directly visiting the site, users entering the site through Google Search, users visiting through a social network, etc) was the same as in other countries. This meant we suddenly registered millions of visits from different sources within the seventh largest city in India!
We were baffled; we couldn’t understand what was going on. After looking over behaviour reports, we found another aspect that caught our attention: the report showing the devices from which this connections were being made.
This information was key to understand what was going on. Against what usually happens, most users connecting from India were doing it from a mobile device. Why now? Who were they? I realized that there was something that was not right, so I began to check the browsers from which these connections were being made. And we found the answer:
Opera Mini was responsible for almost all the visits that were being registered. Let’s take a look at Opera Mini. This browser dominated mobile navigation before Android and Iphone became massive. According to Wikipedia, Opera mini had the capacity to enter sites with a much smaller compress rate.
“The functionality of Opera Mini mode is somewhat different from that of a conventional Web browser, with the amount of data which has to be transferred much reduced, but with some loss to functionality. Most versions of Opera Mini only work in this mode, but the Opera Mini 8 program for iOS, but not the Blackberry/J2ME version, was completely redesigned and can switch between Mini, Turbo, and uncompressed modes, gaining functionality at the cost of lower compression in non-Mini modes.”
Opera has servers that load the sites, when we enter a URL on the browser, that request is sent to the Opera Mini Server, who loads the site and reduces it for comfort in the mobile, optimized it sends it back to the device.
What had happened that specific day? Opera mini had begun to send the devices geographic location as the location of its servers that would optimize the sites before sending it back to the users device. Because of this, users that were entering the site from different countries were being registered in Google Analytics as users connecting from India. Thanks to in depth data analyses, we could rule out other answers and finally we were able to obtain a graph that would prove this theory.
In an almost symmetrical way, the visits that were coming from users browsing with Opera Mini without country precedence specification were replaced for visitors registered in India.