A few days ago I tore into the Gartner Magic Quadrant for Data Warehouse DBMS. Well, the 2009 Gartner Magic Quadrant for Business Intelligence Platforms is out too. Unlike the data warehouse MQ, Gartner’s BI MQ clusters its “Leaders” together tightly. But while less bold, the Business Intelligence Magic Quadrant’s claims are just as questionable as those in data warehousing.
February, 2011 edit: Here’s a partial link that works right now.
Of course, some parts do make sense. E.g.:
- Business Objects’ completeness of vision seems to have been downgraded because of its new affiliation with SAP’s ever-confused Netweaver strategy.
- Microsoft’s completeness of vision is dinged for — well, for not being very complete.
- SAS, which unlike other vendors actually gets customers to integrate BI and predictive analytics, gets top marks in “completeness of vision”.
- IBM/Cognos leads the way overall.
Parts I find more dubious include:
- Whether or not vendors have strong international sales presences affects their “completeness of vision” scores. Huh?
- In-memory analytics are hugely emphasized, to the point that TIBCO Spotfire gets very high “completeness of vision” scores despite being just a portion of an overall BI product line. Yet vendors who get similar performance from allowing drilldown within reports don’t seem to get the same credit.
- Endeca isn’t included, while Spotfire is.
- Despite criticizing Microsoft for not delivering on promised products and Oracle for not doing much at all, Gartner gives both better “ability to execute” marks than are given to Information Builders and Microstrategy.
- While Gartner correctly points out in the commentary that company size is not a strong indicator of ability to execute, this awareness doesn’t seem to have been reflected in the actual chart.
- Gartner’s supposedly rigorous numbers seem sloppy. LogiXML is seemingly cited as almost making the $20 million product revenue cutoff, despite being a company with $7.3 million in overall revenue.
I don’t doubt that Gartner has done good research in support of this article. Indeed, I learned things from reading the supporting commentary. But the actual Magic Quadrant presentation methodology is, as always, fatally flawed.