QlikView 9.0 Reaches for Broader Business Intelligence Market

QlikTech released version 9 of its QlikView business intelligence software today. The product has been in public beta for several months, so the general features are well known to people who care about such things.

Probably the item that attracted the most advance attention is an iPhone version that supports interactive analysis; this also works for other Java Mobile clients like Blackberry. It's cool (or ‘qool’, if you must) but not so important in the grand scheme of things. More significant changes include:

- availability through the Amazon Elastic Compute Cloud (EC2), which lets companies order up a QlikView-equipped server in minutes. (Of course, they still have to purchase a QlikView license.) Users can also expand or reduce the number of servers to match fluctuating needs. Advantages including avoiding the wait for new hardware, no need to physically install a server, and the ability to meet peak demands without making a fixed investment.

- API for real-time updates of in-memory data. This is an extension of previous changes that allowed incremental batch updates and manual data entry. But it still marks a major step towards letting QlikView run time-critical applications such as stock trade analysis, pricing and inventory management. No one will be processing orders on QlikView (hmm, never say never), but the line between analytical and transaction databases just got that much thinner.

- enhanced support for enterprise-level deployments. This includes centralized control panels for multiple servers; load balancing and fail-over; better thin-client support; multi-billion-row data sets; and more efficient calculations. These are critical as QlikView moves from being a departmental solution run primarily by business analysts to a mission-critical system backed by corporate IT.

- free Personal Edition with full development capabilities. The main limit vs. the licensed version is that Personal Edition cannot read QlikView files developed on any other copy of the software, and no one else can read files that Personal Editon generates. The goal is to make it easier for users to try the system on their own – a continuation of the company's long-standing "seeing is believing" strategy.

- functional enhancements including improved visualization, search and automation functions. These are nice but none seemed especially exciting. Changes in previous recent releases, such as set analysis (simultaneously comparing two sets of selected records) were more fundamental. Remember, we're talking about version 9: the system is already quite polished.

Of all these items, the one I found most thought-provoking was the free Personal Edition, which replaces a 15-day free trial. Removing the time limit let users build QlikView into their regular work. The strategy makes sense, but it doesn’t lower the $30,000 - $50,000 investment required for the smallest licensed QlikView installation. Few analysts, who are the most likely users for Personal Edition, have the clout to sponsor so large an investment. Competing analyst tools such as LyzaSoft, ADVIZOR Solutions and Tableau can generally provide a 5-10 user departmental deployment for under $10,000. Although QlikView is vastly more powerful than the others, the lower cost will give them an initial advantage. And once they’re in place, it’s hard to get a company to switch.

On the other hand, maybe QlikView is really moving to compete with traditional business intelligence tools like Cognos, Business Objects and MicroStrategy. QikView’s entry cost is vastly lower than those products, especially once you consider the savings in labor. But most enterprises have a BI tool already in place, so it’s not a matter of comparing entry costs. Rather, the choice is entry cost for QlikView vs. incremental deployment cost on the incumbent. The labor savings with QlikView are so great that it will still be cheaper for many projects. But QlikView will remain be a tough sell because IT departments are reluctant to invest in the staff training needed to support an additional tool.

QlikView will never fully replace the traditional data warehouse and BI tools because its in-memory approach limits the size of its databases. With 64 bit systems, the product can easily handle dozens of gigabytes of data. This is quite a lot, but even the smallest enterprise data warehouses now hold multiple terabytes. QlikView works with such systems by executing SQL queries against them, pulling down limited data sets, loading these into memory, and analyzing them. That’s an excellent and perfectly viable approach, but it does rely on the warehouse being there in the first place.

None of this is to suggest that QlikView has anything but a very bright future. When I first spoke with the company in 2005, it had just reached 2,000 clients; at last count, it had over 11,000. Revenue for 2008 was $120 million and had risen 50% from the previous year. The product has finally attracted attention from analyst firms like Gartner and Aberdeen and is very well rated in Nigel Pendse’s latest BI Survey. My brief fling as a VAR ended two years ago, but I still use it personally for any non-trivial data analysis work and remain absurdly pleased with the results. I won’t say QlikView is better than sex, but its pleasures are equally difficult to describe to the uninitiated. Anyone interested in BI software who hasn’t given it a try (QlikView, not sex) should download a copy and see what they’ve been missing.