The real value of a website may lie in the data itself

IBM Cooperative Solutions Division social analysis strategist Mary Wallace

In a situation where social media sites are gathering more and more data, they may be able to find better ways to use the data to make a profit, instead of advertising as the main way to raise revenue. While the data will allow social media sites to  they need to provide more targeted advertising--as far as I can see, social media sites do have a way of doing this--but the real value of these sites may be in the data itself.

This, of course, is premised on the fact that social media sites are starting to sell data to the highest bidder and find ways to strip out personal identifiers, a move that could lead to questions about data ownership and privacy.

Mary Wallace (Marie Wallace) is a social analysis strategist at IBM's Cooperative Solutions division, who has been at IBM for more than 10 years and is responsible for content analysis. As for Wallace's work experience, she is in a unique position to answer questions about big data, social media, and analytics.

Here's an excerpt from her recent interview:

Q: The real value of social media may not be in selling ads, but in the data they collect. Why would you think that?

Wallace: The reason why targeted advertising can do so well is that it is consistent with a particular activity and supports it; when I search for information about certain products or services, I am willing to get ads that might help guide my search activities. In a way, advertising is similar to a value-added service, and social search makes it more personalized and relevant, which is why Google has made a lot of investments in it.

The key problem is that in most cases advertising can only work in a search-like environment, but for most social media sites, people use their services not for search, but for communicating with friends and family. In this case, advertising becomes the thing that disturbs their communication, and often has some kind of aggression. And the mobile business is widening this situation because, in the mobile world, the limited "real estate" makes the ads look more unpleasant, distracting the user and making the screen messy. Social search can be used as an example of services built on top of social data, but social data can also drive a whole range of services-from market research to consumer/brand engagement, social referral systems, information filtering and professional positioning.

Q: It's one thing to realize the value of data, but how can you refine that value?

Wallace: Refining value from data requires a set of fully descriptive situations, and a clear understanding of what facts will be considered valuable for those situations. For example, when looking for a job, people will have a very clear set of questions to ask and answer: Employee sentiment, business success (revenue, customer, product, growth, etc.), location, demographics, technology, industry, skills, competitors, values, culture, and so on.

In each of the different contexts, the problem can be very different, and so is the analysis associated with it. For example, when people decide where to go on vacation, their concerns are likely to be places, activities, accommodations, weather, costs, demographics, and visitor sentiment. The problem here is that analysis not only needs to identify a field, but also to identify a situation, which is why directed professional services such as LinkedIn or TripAdvisor have been able to deliver higher analytical value to the particular context that they support.

Q: People are concerned that social networks misuse data from user contributions. Is there a reliable way to make data anonymous and deliver data in an aggregated form that strips out individual user information?

Wallace: I think the question of privacy is a more complicated issue. While anonymous user information is part of the solution, I don't think it's the core of the problem. I think the main challenges facing social media in the future will be licensing, trust and transparency. People need to know exactly how their data is being used, allowing social media to use its data for that purpose and only for that purpose. If I have a Tesco discount card and trust Tesco to respect my data, then I would be happy to let them see what I'm doing on Facebook, so they can give me a more relevant special recommendation. Or if I sign up for LinkedIn and I know my data will be available to recruiters and companies that want to hire employees, but I definitely don't want them to use my data for any other unspecified purpose.

There is also the possibility that we will see the presence of information brokers who will provide some degree of indirection (perhaps even obfuscation or anonymity) to represent our middleman. This simplifies the licensing model, but only if we trust the information brokers and trust the patterns they use to control our information being captured by others.

Q: Have the tools been able to match the number and diversity of data so that services such as social networks can start manipulating the data they collect?

Wallace: I've been working on content analysis and semantic technology for the past 10 years, which makes me confidently say that many of the tools we need are ready years ago, waiting for the day when demand will catch up with the supply. The advent of social media, coupled with the growth of a new generation of large data platforms, means that these tools now have the perfect business problems, datasets, and execution platforms to shine their light on. But I think there is a significant technical gap in the industry that will affect the value we get from social networks.

I think the technical challenge that social networking will face is that we deal with large-scale networks because we are moving quickly in the direction of massive graphs with social, semantic, temporal, and spatial characteristics, and we are looking to apply complex analysis to all of these networks. Many of the existing technologies from this interconnected data world may shift to fill this gap, or there will be a new generation of chart databases and analytic algorithms that focus on solving this particular problem. Only time can tell us which technologies will be the winners.

Q: Can you imagine what social networking sites can do for their data?

Wallace: In the medium term, I think we will continue to see that the driving force of social analysis comes from marketing, sales, and support organizations. Social data will be used for market research to help expand sales channels and improve the interaction between brands and customers.

As we move from marketing to sales and even support, the types of analysis will become more complex, which will put pressure on algorithms used to evaluate data and derive insightful insights, identity and presence resolving, micro-segmentation, impact analysis, emotions, intentions, network flow and community momentum, and so forth. More and more social applications will emerge, each of which will deliver market-segment value to consumers, generating professional data for the brand. The social networking ecosystem will promote consumer branding – everything from consumer feedback systems and customer support to product and service innovation. Brand will shift from passive listening/monitoring to focus on active participation, which will require a series of more extensive analysis to optimize those interactive activities and make them operable.

If you want to make a further outlook, then I think personalized expansion can be achieved. Social data is becoming more and more important to personalize each search and navigation experience from Google, Amazon, Netflix and Expedia, but search is only the tip of the iceberg. I expect that, in the longer term, social data will be used to personalize a whole range of user experiences across all aspects of the physical/digital divide, such as the way we shop, the way we think, the way we learn, and in the end we will change our way of life.

You can imagine what happens when we can use data networks to penetrate social networks and semantic networks. Then we will really see that personalization appears in a completely new form.

(Responsible editor: The good of the Legacy)