Given the consolidation in the business intelligence sector, and the recent share price dips even of leaders Cognos and Business Objects, you might wonder why anyone would bring out a new BI product. Certainly there is no shortage of reporting tools, data mining has yet to break out of its statistician niche, while visualisation tools have again failed to become a mass market. However one area that does make sense for a new entrant to focus on is real-time event monitoring, which is typically today addressed (poorly) by the vendors of major applications.
SeeWhy software is a UK start-up which has managed to get over the key hurdle of signing up initial high class customers such as Diageo. It is run by Charles Nichols, previously an executive of Business Objects. Charles is a smart guy who understands the space well. The software pulls data out of real-time message queues, enabling alerts to be generated e.g. for supply chain data in the case of Diageo. The company should continue to focus on this niche in my view, and ovoid trying to be “all things to all men”. For example it would be natural to extend its capability to data mining in order to spot anomalies or trends, but would be wise to partner with existing data mining tools in order to do this. Similarly, if they start to build up repository capabilities and looking at trends in their customer data they should avoid trying to compete with general purpose data warehouse technology, or they risk undermining their message of “real time” analysis. I have written elsewhere how EII vendors struggle when they try and position themselves as general purpose business intelligence tools, since fundamental issues like data quality get in the way if you do not have a persistent store of data such as a data warehouse. This has led to pioneer EII vendor Metamatrix stalling in the market, with virtually no growth in revenues last year. By concentrating on drawing data from real time message queues, marketing to that niche and by selective partnering in other areas SeeWhy should be able to prosper in an apparently crowded market.
I couldn’t agree more. At Stanford we teach that when data mining to be really effective, the data needs to be able to be segmented in many different ways to be able to draw the best actionable results.
To broad and you can find yourself drowning in data and at best treading water.