Business Intelligence is often lumped together as a single market, but in fact there are rather different communities of business users. Many people require some fairly basic information on a regular basis e.g. "what is the sales forecast this month", or "what was last quarter's profitability by product". Despite some BI vendors' fevered fantasies, most people working in large companies do not require a full-blown analytical environment, because analysing data is not their job. However there is a clear subset of business users with job titles like "marketing analyst" or "business analyst" whose job it is to do genuine analysis of data. Whilst they may get some of their data from an enterprise data warehouse, in real life they will tend to also need to combine this with data from a variety of specialist systems e.g. external data feeds or some other applications.
Such analysts tend to have a love-hate relationship with the IT department. They rely on IT to give them access to some of the data that they need, but resent having to wait in line for IT to get around to granting them this access. IT people in turn are nervous of granting broad access to their systems to analysts, who are often very IT-literate but know "just enough to be dangerous" when it comes to data; they suspect that analysts will kick off some wild, uncontrolled trawl through the corporate systems and cause the lights in the data centre to go dim.
Consequently, analysts typically deal with the IT groups as rarely as they are able to, and grab the data they need on an occasional basis before doing what they really need to do using the most widespread BI tool of all: Excel. Excel's great flexibility comes at the price of allowing uncontrolled copies of data to move around the organisation, often without it being clear where the data originally came from and under what assumptions it is based: was this spreadsheet based on data from last month's sales figures, or the one before, and which other data was it combined with? Exactly what definition of "margin" is it using? Philip Howard has written extensively about this problem.
One vendor hoping to address this market of frustrated analysts is Lyzasoft, who, in late September, launch a desktop analysis product (Lyza) following a beta program with over one hundred analysts. This product comes out of nine years of working with analysts through the company's prior venture, Eyeris, a hosted data warehouse offering, which has carved out a small but very profitable niche amongst a number of large corporations. Lyza has a columnar storage system well suited to analytic needs, with a graphical environment allowing analysts to build up analysis iteratively. It has OBDC and JDBC connectors allowing access to most data sources, and keeps careful track of the various steps of analysis that take place, giving auditability. When a user shares results with another Lyza user, all the metadata associated with the result set goes with the results, minimising the problem of how to understand the assumptions behind the data.
The company has a nice series of video demos on their website, allowing prospective customers to get a good idea as to their value proposition. Their challenge will be in overcoming the resistance of IT departments to the introduction of yet another analytic tool (many companies have invested significant effort on reducing BI tool proliferation in recent years). If they can, in a future version, demonstrate a collaborative environment that allows better control of data sharing, then this problem could be turned on its head, and may actually represent an opportunity for hard-pressed IT departments. For now they need to build up a good reputation amongst their target community, and demonstrate that Lyza's intuitive interface is enough to wean analysts off their beloved Excel pivot tables.