A recent article published by the Motley Fool, states that a partnership of International Business Machines (IBM) and Facebook (FB) would be a match made in heaven:

As the Watson Group gets its feet wet, IBM is focusing its cognitive computing efforts on the health care and financial arenas. But transitioning to other industries -- say, for example, the world's largest social media network that houses billions of data bits -- isn't much of a stretch. If Facebook advertisers are enamored with their marketing results now, just imagine combining the power of Watson with Facebook's already impressive targeting capabilities.

With the amount of user data Facebook collects and utilizes, combined with the computing power, predictive capabilities, and learning algorithms of Watson, the results would be mind-boggling.

I'm not going to disagree that Watson's capabilities could be applicable to Facebook, and in a meaningful way. I have to disagree with the way the author of this article portrays a potential partnership.

IBM's expertise in other areas would add more meaningful value to Facebook. Plus, IBM's software solution through Watson is more a marketing moniker, meant to highlight IBM's cloud and analytic capabilities when combined. It's basically software in the cloud, where you upload a ton of data and let the software reason everything out.

What's Watson?

Watson is a powerful computer armed with software that aggregates data and correlates the data into specific meaning. It establishes trends in data, and with that, data is able to arrive at a rational conclusion.

If data can be plotted, and there's high enough correlation between two data points, it's considered to be useful (we measure this with the correlation coefficient). However, data on the scale of Facebook requires a really fast computer to process the disaggregated information into a meaningful trend. When you input data that includes costs, or time, you can assess the rising or falling cost of housing data. Perhaps, you can identify the effectiveness of one form of advertising compared to the other. You may be able to use data to identify when users are most active with pictures, versus status messages. Peak times of day for advertising may be identified by finding the point at which users are more likely to read content rather than post content.

In a sense, if you have a lot of data, you can use standard mathematical principles to identify correlation, and in turn, identify patterns (hence the term pattern recognition). Big data is about finding data points that can lead to a meaningful conclusion about a business. So since we understand the basic premise behind big data, let's look at it in light of Facebook, and how it would work in the partnership ecosystem it already operates under.

Facebook doesn't need Watson; it needs something else from IBM

The computing and cloud aspect of Watson isn't an appropriate solution for Facebook. Facebook has a lot of resources already aimed at data-mining. Remember, Watson's value comes from its software. It doesn't come from hardware. If we were to measure a computer's usefulness based on the amount of clock cycle readily available to it, Facebook's data-center would out-class Watson in terms of computing power.

(click to enlarge)

Source: IBM

If anything, Facebook could use IBM's infrastructure protection and data protection. I mention infrastructure protection and data protection, because the NSA (National Security Agency) has been using Facebook data to track criminal behavior. Granted, Mark Zuckerberg has been a critic of this behavior, and while Mark tries to exert political muscle, maybe a better alternative is to take advantage of IBM's software. Who on planet earth knows better than IBM when it comes to data security at data centers?

Now if IBM doesn't have a very useful solution to Facebook's growing security issues, I'd question IBM's expertise with data security management. I'd also begin to wonder if there's any company on planet earth that could tackle data security when it reaches the scale of Facebook, Google (GOOG), Amazon (AMZN), Baidu (BIDU), etc. There's real market potential here for IBM, and I'm hoping that IBM exploits this specific opportunity in the data center space.

Back to social networking

To understand IBM's role, you have to understand digital marketing, the ecosystem, and industry-specific vernacular. In this case, Facebook is the sell-side of ads, whereas advertisers are the buy-side. Since, Facebook is a platform unto itself, and doesn't aggregate ad-space like the Yahoo (YHOO) Bing ad network, Facebook's sell-side interacts directly with advertisers. Facebook's ad environment is friendly to the vast majority of advertisers, as it is neutral territory. To put this in context, political blogs that primarily lean right, wouldn't be a very friendly ad environment for a politician who represents the Democratic Party. When you go through political elections, Democrats use Facebook's innate data to win swing voters (independents), and Republicans use Facebook's innate data to do the same. Facebook has taken care of many of the qualitative aspects of what makes an advertising platform good. Also, Facebook has plenty of people working on analytics and data analysis. Therefore, I couldn't imagine Watson's innate software adding any practical value to Facebook.

The most practical way for IBM to work its way into Facebook's business model is for Facebook to offer IBM analytics to the buy-side. The buy-side would install IBM's analytics on its own servers, and assess the incoming visitor flow, to determine the overall effectiveness of advertising from the context of internal managerial sales data. Advertisers do not share this kind of information with the publisher (Facebook). So there would be no point in Facebook designing analytic software for advertising partners.

A third-party that's neutral would have to serve as an intermediary that analyzes the inflow of visitors from Facebook's ads and assess the overall effectiveness. IBM's analytic software may result in better targeting and budgeting of advertising dollars, which would improve the ROI for advertisers. Since IBM only provides software, data can be kept in secrecy, and stored on the dedicated cloud.

Assuming advertisers are able to generate efficiency gains from IBM analytics, higher click-through rates for pre-existing ads may result. Since Facebook has no intention of creating more ad slots, the quality of ads needs to improve for Facebook to earn more money from its platform. Therefore, for IBM to be an effective partner it would need to improve the overall effectiveness of advertisers, rather than maximize Facebook's data-mining capabilities.

Conclusion

I think big data has a lot of potential. In this specific use case, I think it's the buy-side rather than sell-side of ads that would benefit the most. Remember, the sell-side (Facebook) earns money on a performance basis. By improving the advertising performance of the buy-side, the sell-side also benefits, making it a win-win proposition.

Facebook isn't lacking in data analytics, and has plenty of data analysts; I think the bigger problem that Facebook faces is in improving the ads that are displayed on its platform. This is something that Facebook has limited control over. If Facebook initiates a partnership with IBM, I'm hoping that Facebook works with the buyer of ads to make them more successful. By Facebook helping advertisers earn more revenue through better analytics, advertisers will advertise more, and in turn, spend more on click-based and impression-based ads.