What Do People Ask Their Social Networks?


We’ve all seen people do it in our social networks, and there’s a good chance we’ve probably done it ourselves now and then. What is it? Asking questions of the people who follow us on Twitter, or our Facebook friends. You know the kind: What cellphone should I buy, what did you think of a particular movie, where can I get rid of an unlicensed handgun, that kind of thing (okay, that last one was me). But what are the most popular kinds of questions we ask our friends and followers? A new study from MIT and Microsoft Research took a look at that by asking a group of users what kinds of questions they asked their networks.

According to the survey, questions ranged from the somewhat open-ended and philosophical (“Why are men so stupid?”) to the explicitly practical (“Point-and-shoot camera just died — need to replace it today for vacation. What should I buy?). The most popular question types were recommendation and opinion questions, such as “I’m building a new playlist — any ideas for good running songs?,” followed by factual knowledge and rhetorical types of questions (we’re assuming that the “Why are men so stupid?” one probably fell into the latter category).

Not surprisingly, technology-related questions were the largest single topic (“Anyone know a good Windows 6 phone that won’t break the bank?”) followed by entertainment-related queries (“Was seeing Up in the theater worth the money?”). People also asked a lot of questions related to home and the family, including “What is the going rate for the tooth fairy?” and questions relating to shopping, travel, restaurants, etc. were also popular, including “What’s a good Mother’s Day gift?”

For the survey, Meredith Ringel Morris and Jaime Teevan from Microsoft Research and Katrina Panovich from MIT asked 624 users, 25 percent of whom were female. More than 70 percent were full-time employees of Microsoft and 27 percent were university students working as summer interns at the software company. Almost half of those surveyed were between 26 and 35 years old, while almost 30 percent were between 18 and 25 years of age, and 25 percent were 36 to 45 years old.

The survey showed that almost all of the users who responded (98 percent) had Facebook accounts, and 71 percent had Twitter accounts. The average size of a social network on Facebook was 209 friends, which is interesting if you’ve ever heard of Dunbar’s number — a theoretical limit on the number of people the average person can remain connected to in any meaningful way, which the social scientist said was usually 150. Most users who took part in the survey said that they asked their social networks questions because “I trust my friends more than I trust strangers,” while the second-largest response was “A search engine can provide data but not an opinion.”

Interestingly enough, as part of the survey the MIT/Microsoft team asked respondents how many times they updated their Facebook status or their Twitter page, and found that more than 21 percent of those surveyed — the largest single group — updated their Twitter page a few times a day, whereas only 10 percent said the same thing of Facebook. The most common response for Facebook came from 30 percent of those surveyed, who said they updated their status a few times a week.

Although drawing a lot of major conclusions from this survey wouldn’t be wise, considering its small sample size and the fact that it was composed entirely of Microsoft staff, there is no question (pun intended) that social networks are increasingly becoming the place where people find answers to their questions, from the trivial to the serious. The fact that this is happening, and the fact that many of those questions involve purchasing behavior, has implications for businesses and also for search giants like Google and Microsoft. For more on that, check out what Om had to say in a GigaOm Pro report called Why Google Should Fear The Social Web (subscription required).

Post and thumbnail photos courtesy of Flickr user Oberazzi


Taylor Marek

Very interesting survey. What I find most interesting is the top 29% of questions are for recommendations. Add in the person who said, “I trust my friends more than I trust strangers” and you have the reason why social media has revolutionized everything we do online and off. Businesses should take a good look at this survey, it gives all the answers.

Phil Hendrix, Ph.D.

Thanks, Mathew, for sharing this interesting study and results. I would also recommend the paper – available at http://bit.ly/8UdMkc – to interested readers, as it contains additional data and a description of the sample, methodology, etc.

I think the researchers make two contributions that are particularly noteworthy: (i) they develop a classification of query types and topics based on a content analysis of some 249 actual queries, which is useful in and of itself; and (ii) they explore users’ motivations for using social media instead of a search engine. It will be interesting to see how Social Media sites and Search Engines compete over time to deliver “opinions I can trust.” As Marshall points out, Hunch (www.hunch.com) is trying to bridge the two.

I would raise a couple of cautions re: the research methodology that readers should keep in mind:
1. As the researchers acknowledge, the sample (all MS employees or interns) is not representative of social media users in general. Moreover, women are underrepresented (25% F/75% M) in the sample, compared to distributions found in studies based on much larger samples – Pingdom (http://bit.ly/6P2LL2), for example, finds that Facebook and Twitter skew toward female (60% F/40% M).
2. Results should also be interpreted with caution due to the “self-reported” measures of frequency. Again, as the researchers acknowledge, self-reported data can be inaccurate – in our experience, recall of frequency (“how often do you…”) are often imprecise. So, the relative frequencies reported in the study are probably best viewed as directional.

Both of these difficulties could be overcome with a research approach that uses a Text Analytics tool such as Lexalytics (see http://bit.ly/3Vd4Oi – “Twitter? Yes, we can analyze that”) to examine a large sample of actual posts. This approach offers several benefits:
1. Robust Sample – can examine a large number of posts (10-20k), drawn from a large number of social media sites (15-20), on different days, at different times of day, etc.
2. Robust classification – posts can be classified and subsequently analyzed based on numerous criteria, including type (update; query; answer; etc.); topic; day of week; time of day; even sentiment, all by particular site
3. Repeatable – the analysis could also be automated and repeated, allowing one to examine how posts are changing over time (both in the aggregate and by site)

Chris D

“there is no question (pun intended) that social networks are increasingly becoming the place where people find answers to their questions”

Help me out here, because I think this is still a question. The article seemed to be all about social networks being the place people asked questions and not the place questions were answered which I think is much more important. Are we supposed to presume that questions asked = questions answered? What types of questions people respond to is much more important to me.

Sanjay Maharaj

These questions present a real opportunity for Twitter to monetize the Tweets. When will we start seeing Google stlye Ad Sense on Twitter? This is a great treat to Google


What’s interesting about this is how much people might not even realize the gravity of asking their social graph a question like “what kind of camera should I buy?”

What’s even more interesting would be to look at the recommendations that are given by their friends, and compare their answers to against the latest campaigns by those companies and see which campaign was more memorable…

Scott Hoffman

Mathew – great data here – thanks for finding it!

Did the study show what % of the total status updates were questions for any particular user? For instance if someone updated their status a few times a week, how many of those updates were questions?

Comments are closed.