Twitter recently released detailed data about when people around the world sleep -- mostly at night, as it turns out.
But what the social media giant was really demonstrating was its ability to reveal subtle insights into the daily habits of millions of its users around the world – and to help brands harness that extremely valuable information.
Two of Twitter's data geeks, research scientist Jeremy Lin and analytics and data visualization expert Miguel Rios, looked at tweet patterns in four cities: New York, Tokyo, Istanbul and Sao Paulo, then shared their findings on Twitter's engineering blog.
Essentially, the patterns show what many people would consider obvious -- that generally speaking, Twitter users tweet more during the day, than at night when they're sleeping.
There were some variations in the Twitter "heatmaps" for each city that Lin and Boricua released, but generally the revelations were pretty obvious.
In Tokyo, waking/sleeping patterns were relatively consistent throughout the year, while the other three cities experienced seasonal variations.
And in Japan, people tend to post more tweets in the evening than they do during the day, compared to the other cities which showed more usage during the day.
And in Sao Paulo, the volume of tweets goes down in the afternoon, when many people take an afternoon siesta, but the nights there get shorter in August when many Brazilians are out later and stay awake longer.
While the data itself may not be Earth-shattering, it demonstrates Twitter's unique ability to collect precise information about its 140 million users around the world, says Neil Bearse, associate director of marketing at Queen's School of Business.
"Data surrounding when people are awake and when they're asleep might not be the most useful type of data but I think what they're demonstrating is the collective power you can get out of millions of tiny data points," he told CTVNews.ca.
While traditional marketing surveys are costly, time consuming and often easily skewed based on the type of questions and how they are asked, the new Twitter data is relatively easy to get a hold of and is purely a function of how people use the service -- not how they say they use the service.
Bearse says the release of Twitter's data comes just a few weeks after General Motors announced it was backing away from pursuing Facebook ads, after having determined they had little impact on customers' car choices.
That decision caused a tremor to ripple through the rest of the social media world and may have prompted Twitter to release the new insights.
"I think there's a general market sentiment that's questioning the value of the mass of data that social media sites are collecting," Bearse said.
"For the longest time it's been considered this treasure trove of demographic information and recently with GM announcing they're no longer pursuing Facebook ads, Twitter might be saying 'look at the value of our data, look at what we can do with it,' specifically to send a message to brands."
Deep insights about when people are sleeping and when they're awake are unlikely to shake up the advertising world, Bearse acknowledged. But similar data visualizations, tailor-made to target specific groups, could be incredibly valuable to brands looking to reach very certain demographics.
For instance, Twitter could identify a group of users interested in running, based on who they follow, what they tweet about, what they retweet and their biographical information.
They could then track that group's habits and propose key advertising windows to Nike or New Balance based on when those runners are most likely to be scanning their Twitter feed.
Hypothetically, Bearse said, "Nike might want to run ads between 6 a.m. and 10 a.m. when Twitter has determined most runners have completed their weekly long run, not between 6 and 8 p.m. when runners normally nap."
As a near-instantaneous measurement of the world's reactions to global events, Twitter has already demonstrated not just the ability to track events and reactions as they occur -- but also to predict trends.
Yes, Twitter will tell you when a home run has been hit in the World Series before it appears on the news, or even when an earthquake has occurred before some people feel the ground shake.
But Twitter has also allowed researchers to predict phenomenon such as influenza outbreaks, before they happen.
A group of researchers based at Southeastern Louisiana University analyzed 500 million Twitter messages over an eight-month period in 2009 and 2010 and were able to forecast future influenza outbreaks using a small number of keywords to track the prevalence of flu-related messages.
Their predictions were accurate 95 per cent of the time, when correlated with health statistics collected by the U.S. Center for Disease Control.
Whether it's analyzing napping patterns or predicting the next major health emergency before it occurs, Bearse said Twitter is on a mission to prove that the ocean of data it collects can be harnessed and monetized, and will ultimately make the company more valuable if and when Twitter decides to go public.
Also, he said, Twitter's army of geeks are just trying to prove to the world that data is, well, cool.
