Analytics FOR THE REST OF US
Perhaps most importantly, whether or not those software companies fulfill all of their promise, they will leave a legacy where opinion in the presence of data won't be tolerated. Edward Snowden may well have exposed less than ethical data collection methods by U.S. government agencies, but he also highlighted something that the U.S. government is rarely accused of: acting on data and reason rather than opinion or ideology.
While you read this article, millions of e-mail messages were sent, hours of video uploaded and thousands of status messages posted. If predictive analytics can find value in all of the data we collect, perhaps it will unleash a new wave of reason-based insights and innovations across multiple industries.
PREDICTIVE ANALYTICS MARKET
Small companies that are starting to attract attention include Ayasdi and Context Relevant. Although Ayasdi recently raised $30.6 million in Series B funding and Context Relevant $7 million in Series A funding, they are not likely to become big companies before they are acquired by the likes of SAP, SAS, Oracle or IBM. And they are not alone in the field. Other companies like Skytree, BigML, Yottamime Analytics and Predixion Software are building similar capabilities. (Download PDF images regarding Ayasdi and Context Relevant-also on page 12 of KMWorld, February 2014, Vol 23., Issue 2).
The differentiators for those firms will be production and results. The firms that build models that can easily be embedded into existing transaction systems and provide ongoing value will be best positioned for success. One-off, batch mode solutions will not work with today's rapid-fire economy.
Value, however, must be sustained. If models prove more fragile in practice than in theory, errors or misinterpretations of results could lead to a backlash against the technology-and such errors don't need to be universal across the vendors. Any major error that becomes a public relations problem for one firm would force others to shift from valuable development to defending the technology and risk mitigation, which is never a good situation for an emerging market.
If, however, multiple companies provide solid, repeatable results across a range of domains, within five years that could lead to a boom market for predictive analytics in terms of startups and M&A activity.
Deployment models will be an important consideration for customers, as the increasingly proprietary nature of data may well make cloud solutions less tenable. The ability to deploy on premise will help ensure that companies can mitigate data security-related risks.