Who has more information about US’s Gen Y? Facebook or the United States Government?
Each time I ask this question on stage, most hands vote that Facebook has more information, although someone usually suggests the United States Governmant can obtain it if they need to.
In fact, the amount of consumer data emerging in the social web only continues to hockey stick (see this graphic from Twitter) The following is a summary of a research project we did to understand how to harness new data types in their online experiences.
The 7 Elements of Social Data
We segmented data that has been tried and true for years on the top tier: demographics and product. Then, we segmented data that digital marketers are striving to tackle now in the middle tier: Pschographic, behaviroial, and refferal data. At the bottom tier, we listed out experimental new types of data that most companies have not harnessed, the newcommers location based and intention data.
Demographic Data
This data types enables an effecient way to create context about consumers, yet broad survey-based research may not yield specific nuances and needs about specific individual taste as today’s consumers are given more choices and have more discrete needs. Some marketers are able to glean demographic data from social accounts gender, age range, by profile data, profile pictures, or searching public records like Zabasearch and Spokeo.Product Data
A data type commonly used in ecommerce websites, this data type is used to match similar products with each other, in order to cross-sell and up-sell products. Often combined with demographic data, this data type, mixed with referral and behaviorial data yields greater accuracy. Visit any ecommerce website from Amazon, BestBuy and beyond to find examples of product matching.Psychographic Data
As the social web exploded in the past few years, consumers are volunteraily self-expressing their woes, pains, and aspirations in websites. This provides those who want to reach them increased opportunities to market based on lifestyle, painpoints, beyond just product sets. This data type is useful in both message and conversation creation as well as identifying features and products to improve or fix. To learn more about lifestyle and pain point positioning see the 5 stages of positioning by Lifestyle, Pain, Brand, Product, or Features.Behavioral Data
There’s at least two ways to find this data, it’s in both existing customer records like CRM or ecommerce systems or also in the “digital breadcrumbs” that users are leaving in social networks using a variety of web techniques from cookies, FB connect, and other social sign on technologies. The opportunity to suggest content, media, deals, and products to them that matches their previous behaviors will yield a greater conversion.Referral Data
Customers are emitting their recommendations for products, but positively –and negatively. Both explicitly through ratings and reviews, as well as implcity though gestures like the ‘like’ button to their social network. Vendors like Bazaarvoice (disclosure: client) offer a suite of tools for customer feedback and intelligene, Zuberance fosters positive WOM through positive ratings, and ExpoTV is a catalyst for conversation using video reviews, and see the well known case study from Levi’s who implemented the Facebook Like button.Location Data
As location based technology and services emerge for consumers to emit signals where they are using mobile devices, this data helps to triangulate context around location and time for brands to reach them. From Foursquare checkins and the associated contextual ads that emerge to ‘players’ to Facebook places, consumers can now emit their location, in exchange for contextual information, see how Awareness Hub (client) is able to surface influencers by location in Foursquare Perspectives.Intention Data
The most innacurate, this volatile data type holds great opportunity to predict what consumers will do in the future. Wish lists, social calanders like Facebook Events, Zvents, and aspirational websites like PlanCast, 43 Things allow consumers to broadcast their future plans
Savvy marketers will harness explicit content and serve up the right messages in advance as well as poach from competitors. Learn more about intention data –which is faster than real time.
Conclusion: Use Data Elements in Combination to Yield a more Potent Elixir
In our initial findings we had many other types of data, but organized these in as discrete buckets as possible. While this seven segmentation types makes sense today, we expect many other types of data to emerge just as technology and consumers do. Be sure to use in combination for reaching maximum potency in your ‘data elixir’ –relying on only one is no longer sufficient.
Want to learn more? I did an hour long webinar on this topic, you can watch the replay at NetBase, who hired me to present this original research. Scott attended the webinar, and reviewed the highlights. Thanks to Altimeter’s Christine Tran for help on this project.
Hi Jeremiah – great post and fantastic distillery of the peta-exa-multi-bytes of info out there. You might add Governmental information as an 8th type. That 8th type could be sub-divided into unique identifiers, property information, tax information, job information, etc…
So – given explicit, implicit and implied U.S. Constitutional protections of privacy and free speech – Here is another interesting question – given the vastness of information that the US Govt has, and the information that a large data-driven company like Facebook or Google has – Should there be any law on the books to keep Govt Agencies from using/accessing/data-mining private company data? If they could combine their 8th type with the 7 you’ve identified… that might be *bad.*
That’s interesting, John. Isn’t some of that government data you list potentially the demographic data? “Government data” could be a subset of this, I think it may be more aptly named “Citizen” data.
I think there are some discussions online about that very question around the Patriot Act.
Citizen data is a much better wording. There is overlap with your 7 categories above, but the government also knows very specific SSN, Drivers license number, Passport number, employment information, tax information, residency, social security information, veteran information, etc. If they had what they currently have, and additionally also had a Facebook-Google depth of your 7 categories…
Agreed, this is the amazing part of this, as soon as you tap 2-3 of the elements it’s easier to start unfolding this story. I know there’s overlap, it was a very difficult task to try to segment, but after much deliberation we arrived at these 7. More will emerge later.
Agreed, this is the amazing part of this, as soon as you tap 2-3 of the elements it’s easier to start unfolding this story. I know there’s overlap, it was a very difficult task to try to segment, but after much deliberation we arrived at these 7. More will emerge later.
Curious what happened to technographics? Certainly they can be found in behavior, but based on past research from you, Charlene and others, I’ve always thought the technographic profile of an audience was critical in assessing the social elements of your marketing efforts. Where does that fit?
Curious what happened to technographics? Certainly they can be found in behavior, but based on past research from you, Charlene and others, I’ve always thought the technographic profile of an audience was critical in assessing the social elements of your marketing efforts. Where does that fit?
I’d agree with Jason on the Technographic comment. And I guess the bottom two are breakouts of “Socialgraphic”? That is another data set term I have heard used.
I do see great possibilities with both Location and Intention Data, but the reaction time might be simply too fast to capitalize on?
We don’t do “Technographics”, at Altimeter, we’re focused on Socialgraphics. We have a report coming live on that topic soon, which of course ties into mobile and beyond. These 7 layers don’t conflict with our focus on Socialgraphics –its just another way to look at it.
We don’t do “Technographics”, at Altimeter, we’re focused on Socialgraphics. We have a report coming live on that topic soon, which of course ties into mobile and beyond. These 7 layers don’t conflict with our focus on Socialgraphics –its just another way to look at it.
We don’t do “Technographics”, at Altimeter, we’re focused on Socialgraphics. We have a report coming live on that topic soon, which of course ties into mobile and beyond. These 7 layers don’t conflict with our focus on Socialgraphics –its just another way to look at it.
Synthesizing these elements requires analytics that have not yet reached maturity. There is an explosion of semantic (sic “artificial intelligence” or “machine learning”) technologies that are just now coming to market.
The advice in this posting is hard to execute upon until there is a way to synthesize all of these data feeds using contextual intelligence. And that, my friends, will change everything.
Those 7 elements is only for wide view, you need to view more detail on it because those 7 elements still able to break to lot of small part. To know your customers is not easy.
Hey Jeremiah..
Good breakdown of the data categories. One could argue marketers, content strategist and experience designers have been planning against these same dimensions long before social.
What would be an interesting analysis is the underlying data or meta data to in each of these categories. The powerful elixir as you state comes when you can connect these data elements together. At this high level it may still be hard for people to do this data mapping.
As you have stated some of these categories are “data” or information like products while others have more implied logic like behavior and preference. The “logic” combined with the “information” increases relevance of experience.. This was a similar promise of personalization.. Which is another conversation..
Hope all is well..
dirk
There will never be a true solution that delivers empirical reports that turn data into insights and insights into ideas that can then be acted upon with a push of a button…we have been waiting for AI to mature for decades and t may never get there. Relying strictly on NLP type technologies will limited your ability to provide value today. Our SMMS uses a combination of human and readily available technologies that blend traditional primary research methodologies with newer, more novel semantic search algorithms. The results are pretty compelling for our value chain clients. What Jeremiah outlines here as well in earlier reports like their “Framework for Social Analytics” helps us develop our own framework and methodology for delivering value today, rather than hope for a true solution that integrates and synthesizes misspelled, untagged, 140 characters check-ins and tips. It’s an imperfect world.
@twitter-27450194:disqus : Your SMMS sounds very interesting. Is there any chance we can see a demo?
We recently combined NLP with machine learning to extract the meaning of text documents as a proof of concept that unsupervised AI can detect the meaning of any document from any perspective (keyword) regardless of spelling mistakes, etc.Â