The flash and splash of last week's SAS Social Media Analytics (SMA) launch belies the apparent very early, incomplete state of the service. The SAS Global Forum launch created inflated expectations -- CMO Jim Davis said "the uptake on [the announcement] has been far greater than we ever expected" -- leading to glowing reviews of a tool that is nowhere in production, that is being moved toward production use by exactly one customer. SAS does have excellent data-mining, text-analytics, and visualization technologies and proven skill at bringing them to market, but let's get real. The dashboard interface is attractive, but the application is nonetheless narrowly focused and inflexible, an analytics silo with no out-of-the-box integration with enterprise systems. It is simply grossly premature to characterize SMA as having "a unique character that stands out in the market as, at this point, one of a kind and thus doesn't have anything to compare against as of yet," as Paul Greenberg did in his SocialCRM blog.There is no question that SAS analytics offers an unmatched combination of depth, breadth, and business utility. SAS has deep technology and an intense customer focus that generated $2.31 billion in 2009 revenues. Add a hot, new information source, social media, and you get a great business case for a new product that will appeal to an unreached market, enterprise public-relations and marketing departments that need to understand perceptions of their corporate personas and work to shape opinion.
SMA is an on-demand, hosted solution. Nothing wrong with that: as-a-service offerings surmount adoption barriers by lowering entry and operational costs and speeding implementations and therefore time to insight. Further, SAS has collected certain news- and social-media feeds over the course of two years (as have, similarly, other social-media analysis providers over varying periods), a valuable asset.
Yet each SMA implementation involves (or will involve) significant customization taking 4-6 weeks, according to product marketing manager John Bastone, and each of the next year's first half-dozen SMA implementations will be unique. Bastone stated that the "on-boarding process" entails identifying concepts and developing taxonomies pertinent to the prospective user's business domain and goals, whether, for example manufacturing users concerned with product design and quality or banking users focused on blogger influence. Implementation runs a minimum of $50,000 for "set-up and ramp-up" with a $10,000 monthly cost.
According to John Bastone, "the use cases are very diverse," and "the end goal is completely different with every client we talk to." SAS developed the current on-demand instances in pilot work for a bank and for a hospitality chain, although only one of the two is moving toward production. "We're being very diligent and focused on the accounts we will sell to in the first year," said Bastone.
So one should not mistake SMA for a generally available product, nor for a do-it-yourself analytics solution. The recorded demo on SAS's Web site shows analytics feed by social- and news-media sources. I see an analytics silo that barely scratches SAS-platform capabilities. For example, I see very limited ability to explore data at a feature level -- to look at sentiment or opinions associated with particular named entities or topics discovered in source content -- SMA's phrase and word clouds appear to be little more than eye candy -- or to compare relative sentiment across sources or topics. (By the way, the feature that allows you to "forecast future social media activity" seems like a joke, premised on predicting what will be hot next year from this year's social-media mentions.) I'd love to see, for instance, the kind of document- and entity-level analytics and visualization that Silobreaker.com has now been delivering for over two years.
Further, I see no capability in SAS SMA to match sentiment trends to content-detected events such as advertising impressions, marketing campaigns, or market news. I accept SAS product marketing manager John Bastone's response, that SMA is backed by the power of the SAS platform, that part of set-up is (or can be) create a custom solution that brings contact-center, CRM, and other corporate data assets into the solution. The need for customized, ad-hoc data and application integration, with data duplication and likely data latency (delay) and on-going professional-services involvement, is not appealing, but without that integration you see, in the words of my Twitter-mate Jim Brewer, "effect without cause."
I'm not the only one with an integration question. In a video recorded at SAS Global Forum, Bastone responds to a question about SMA integration with another SAS On Demand application, Marketing Mix Advisor. There is no integration. Further, it does not appear possible to export SMA outputs exportable into other SAS software components. SAS Social Media Analytics is a siloed application.
SAS Social Media Analytics is a foray for SAS into a new market and a new information source. SMA is based on stand-out technology and corporate capabilities, but folks, it's not All That. SMA is a siloed application, one with no production users that will likely be rolled out to fewer than ten customer sites in the next year according to the company. Next year's SAS Global Forum, April 4-7, 2011 in Las Vegas: that's when we'll see if and how SAS's social-media analysis bets have paid off.
Addendum, April 19, 2010, 10:09 pm: I was remiss in not disclosing that SAS was a paying corporate sponsor of the April 13 Sentiment Analysis Symposium, which I co-own, and that SAS staff bought me dinner the evening before the symposium.The flash and splash of last week's SAS Social Media Analytics (SMA) launch belies the apparent very early, incomplete state of the service. The dashboard interface is attractive, but the application is nonetheless narrowly focused and inflexible, an analytics silo with no out-of-the-box integration with enterprise systems.