Posted by Marshall Sponder on October 29, 2010 | Link It
Occurred to me recently that Google can be, under certain conditions, an excellent tool to categorize sites and media properties with, though I realize it will not work in every case. I’ll talk alot more about this in my book on Social Media Analytics (a new book site is being worked on where I will talk about the book as I’m writing it and ask for feedback).
But here’s one example, if I were to take a list of sites – any list – could be Comscore‘s 50,000 or so sites, or just any list all of blogs, message boards, photo sharing sites, main stream media outlets, whatever, and run a set of pre canned queries on them while counting the results, I could tell you how relevant they were on the subject of that query. Sure, there would be some issues with dynamic urls and sites that are serving up a lot of duplicate content (which Google tries to suppress with the duplicate content filter) but over all, if I have a good set of queries and enough time, I could categorize a bunch of sites with relevancy for a particular subject (what the query is about).
If I had enough different queries, and enough time, I could categorize the web (but right now, without a bit of programming, this would be impossible to scale); in fact much of Comscore is manually deciding, via a dictionary team, what categories a site is in. And Google collects information via Google Analytics Benchmarking where sites that share data can compare themselves to other sites who also share anonymous data in a category (say, magazines) and see how they preform on 6 preset metrics.
Much of this got stimulated by looking at CisionPoint and Recorded Future, two platforms I’m playing with right now and will have more to write about them in, lets say, the near future. I’m also giving a webinar with Jay Krall of Cision in mid November on all the neat things CisionPoint can do and it will be an interactive webinar where I’ll be asking Jay some cool questions and he’ll show people what CisionPoint actually does.
For example, CisionPoint now reads in Radian6 data and merges it with their Media Outlet database and Industry Segmentation – I bet a lot of people didn’t know that or what to do with such information.
Results of the survey are interesting and noteworthy yet I question if 1753 interviews can be extrapolated to map to precise numbers of hundreds of millions of Americans which this study does (that might be standard practice).
Over the last year people who use social networking sites or services frequently (several times a day) increased by 12%; however users who infrequently use social networks (a few times a week) stayed the same. The study makes a claim that seems to contradict point 2 and states 39 million users go to social networking sites several times a day, more than double the 18 million users who did so in 2009. They get the numbers by working off the 1753 interviews and equating it to some the total number of social network users in the US based on panel data that Edison / Arbitron must have based on panel data.
The study says more woman are on Social Networks than men (57% vs. 43%) – again, based on 1753 interviews. I think the percentages are probably not too far off but wonder if the sample size was too small.
Students are most likely, as a group to be frequent social networkers (25%).
I suppose it makes sense that people who are on social networks a lot will also be more aware of brands who are present there.
There’s a lot of mobile behavior stuff in the study including an interesting 55% who have played games on mobile devices – which may explain why Google is getting into Mobile Gaming and Online Gaming, in general.
In fact, frequent social networkers are more likely to credit mobile phones as having the greatest impact on their lives as opposed to the general population overall.
Also interesting is that frequent social networkers are much more likely to give up watching TV than the general population and are far more likely watch TV programming on the Internet (which makes it easier to give up watching TV altogether).
People who are frequent Social Networkers will be twice as likely as the general population to purchase music by digital downloading it and are much less likely to buy music on CDs or not to buy music at all.
One of the main conclusions of the study is Americans who check their social networking sites several times a day are much more likely to be young and female (a conclusion I question). The study also says that frequent social networkers will also be following brands more closely than the average online American who isn’t a frequent social network user.
I wonder if there is a way to test some of these ideas out – I know Comscore‘s Segment Metrix could provide information that might confirm if the information provided by The Social Habit study is as wide spread as it’s asserted to be here – but I don’t have access to Segment Metrix or any of Comscore’s suite of audience measuring products – so I can’t go there.
On the other hand I do have Compete.com and we can look at some sites that are popular social networks or services (like Pandora) and perhaps, see some of the trends mirroring the study – or perhaps, not.
For example, Compete.com says that 55% of Facebook visitors are female – that seems to dovetail with what the study says, but we can’t segment frequent users of anything on Compete vs overall users – so most of the Social Habit survey can’t be validated using tools that are easy to get hold of.
When I tried using Quantcast on Pandora I find only 2% of the visitors where frequent users and they generated 37% of all the visits – but that was far more than regular visitors (38%) that ended up generating 47% of all visits to Pandora.com.
I think the value of studies like this for an analyst like me is to suggest segmentation studies that are possible to do even with Google Analytics – providing one has access to the right sites – something that is very, very rare.
I think the study is interesting and worth writing about – yet answers only part of the puzzle of online behavior of frequent social networkers.
I proposed that Social Media Marketing, augmented by Social Media Monitoring – is really about lists and action, reaching out by the community manager – and here I’m presenting the second part – the first table of1 year of tweets collected by Radian6 for Havana Central Restaurant Chain – I will discuss and further refine this chart in the next post on this subject – but I want my readers to think about what I’m presenting now (you can also download the entire spreadsheet here at case study -2).
This chart doesn’t yet filter on actual check-ins to any of the the restaurant locations – is just looks at all the people who tweeted about Havana Central in some way over the last year and how many times they tweeted each month about us.
I sorted by the top tweeters but there were 960 unique Twitter handles that tweeted about the Havana Central over the last year.
What do you think the reward should be for someone that tweets every month – a free drink – a reach out by the community manager to each account?
Working this list?
Isn’t this where the Social Media ROI really is?
Think about it and please comment – and feel free to download the entire list here - case study -2
Marshall Sponder is an independent Web Analytics and SEO/SEM specialist working in the field of market research, social media, networking and PR. He provides digital data convergence generating ROI and develops data metrics, KPI’s and dashboards that drive businesses by setting, evaluating benchmarks and teaches Analytics at UCI Extension and Social Media for The Arts at Rutgers University.