Introduction
With modern text analytics tools, it is possible to mine social media and summarize what users are saying about Facebook and Google+.
It is also possible to identify:
- users’ stated intentions (like intention to leave)
- trends in the volume of invites being requested and sent out.
Media Impact Graph
The following graph shows the count of published news articles on Google+ between 30th June 2011 and 10th July 2011.
You can see the number of articles increasing steadily for a number of days after the launch. The peak is only reached a good seven days after launch.
This reflects the effectiveness of publicity and this is a very important metric (media impact greatly influences the initial interest in the product).
For a social network to catch on successfully, the initial media impact is extremely important. If people get on a social network and don’t find their friends on it, they are soon likely to abandon it as uninteresting.
So, one key predictor of the success of a new product that depends on network effects is how many news articles appear about the product divided by the time.
The media cascade in this case was 7 days long. So, Google+ seems to have enjoyed a very favourable media curve.
Interestingly, some of the buzz for Google+ would have come from the FB announcement of the tie-up with Skype (it came within the critical week).
Speculation
We have a pet theory (not yet supported by evidence) that a media cascade needs frequently emerging tidbits of information. We think that rumours and information leaks allow late-comers to have something new to report in addition to the main news that has already been reported.
The stream of announcements emerging from/about FB included:
- Stories that emerged about an FB circles hack
- The announcement about the tie-up with Skype
The contribution from Facebook might have been avoidable. Any responses could have been highlighted in the Facebook UI without any reference to Google+. A reassurance to Facebook users could have been provided by other quieter means.
It is possible that FB’s reactions did contribute to Google+ keeping the media spotlight on itself while it mattered.
But of course, the above point is made on the basis of an assumption (that a media cascade needs frequently emerging tidbits of information to feed upon).
Search Engine Counts
The wrong way to go about measuring media impact is to feed the product name into a search engine like Google or Bing and note down the number of search results returned.
When this author performed a search for Google+ on Bing using the Bing API, on July 3rd, he was shocked to see Bing report 168000 results for the keyword Google+. However, when the author skipped to the page that contained the 200th result, Bing reported that the number of results available was only 261.
Google had similar behavior reporting numbers of 9930 on page 1 and 190 on page 19.
This is because search engines sometimes return estimates for these counts that can be way off, and only bother to calculate the real counts when you go down to higher numbered pages.
Customer Impact Graph
The responses to the news that we can track most easily are:
- The volume of Facebook mentions in the public news feed.
- The volume of tweets about the competitor.
The timeline is in Indian standard time.
Notice the daily fluctuations – the peaks occur when it is mid-night in India. That is morning to mid-day in the USA and mid-day to evening in Europe. So twitter seems to have more active users in Europe and North America.
It is informative the compare the two graphs above. Twitter mentions seem to be growing even after news mentions have dropped sharply. It is too early to tell what the cause of this difference is.
It could be a sign that Google+ is sustaining interest, even after new articles about the offering have ceased to appear.
It is also possible that a customer mention graph is just a lagging indicator.
Hourly Mentions on Facebook
Compare with trend on twitter with the trend on Facebook (public feeds).
It’s a lot harder to spot trends here. It’s a problem with some techniques used in text analysis. Some techniques like trend analysis are only possible when there is a lot of textual evidence on the phenomenon you are setting out to study.
There are other methods, like summarization, that can still be used when the volume of data is smaller.
Pre-launch Sentiment Trend
You can find below a few graphs tracking customers’ willingness to try Google+. With social networks, one measure of willingness is the number of explicit statements of interest in a product, as in “I want to try Google+” and another measure is the number of invites requested.
The latter measure yields the following graph:
Note that this graph reflects only invites requested and not invites offered.
It sure seems as if Google hasn’t done a good job of communicating that Google+ is now fully open and doesn’t need an invite to join.
But on the other hand, people sending out invites are good publicity for a social network, and people are more likely to invite their friends if they think their friends can’t get in by themselves.
Critical Sentiment Trend
Critical actions are actions like quitting that are potentially critical to business. We show below a graph of ‘quitting FB’ announcements over time.
On Twitter, the graph looks like this:
Can we draw any conclusions from this graph?
This might mean that after seeing Google+, people are coming to the conclusion that they prefer Facebook, or that they are happy using both.
But we cannot say that for sure without more information. That’s because the ‘quit FB’ mentions might just be following the trend in news coverage.
