A couple of weeks ago salesforce announced that it has acquired predictive analytics company Prior Knowledge. The announcement came on the day before Thanksgiving and hence was little reported – interestingly the Prior Knowledge team moves quickly to shut down the API for their public product, Veritable, as well as permanently delete user data. They even went as far as to quickly shut down the Prior Knowledge website – that’s a pretty unusual move, normally acquired startups at least give users some warning of an impending closure. As one commentator mentioned to me;
"Web site gone. API turned off. Customer data deleted. That’s a pretty abrupt post-acquisition shutdown"
Anyway – the elegance or otherwise of the shutdown aside, this is a pretty interesting acquisition. I’ve long predicted that salesforce would move strongly into the area of analytics, partially to drive more value for its existing products. As I wrote in my pre-DreamForce post this year:
"…I sense that the company is close to getting to the point where the mass of information on individual networks makes it difficult to use the tool in the largest organizations. I’ve previously said that it is through the intelligent use of analytics on Chatter data that Salesforce will be able to surface relevant content in a way which requires little user input. I’ve seen a number of third parties trying to solve this problem and I’d not be surprised to see Salesforce acquire one of these companies and signal its intention to move into this area"
Alex Williams covered the company a few months ago and explained how the technology uses machine learning to build knowledge into software – essentially developers can build apps that can intuit what something is by looking at statistical patterns within the entire dataset. In doing so the service helped to fill in missing values within data sets – by extension it could be used to surface relevant content within a stream (Chatter anyone) to make services more useful. Essentially they’d democratized dataset analysis to drive analytics-type benefits in an automated manner. They create data out of patterns and in doing so provide a really useful service.
Or at least they did – since now the service has been permanently shuttered.
It’s not all bad though, there are a wealth of really interesting problem areas that the Prior Knowledge technology could be applied to within the salesforce stable – Chatter is, as I mentioned, an obvious example, but there are also applications across the entire product line, as well as potentially offering the technology as a service to application developers within Force.com, Heroku or Database.com. This will be an interesting one to watch develop.