Just coming out from a meeting where the clear shift from conventional data warehouses is being applied to DW world. As the technology is changing so is the interpretation of DW itself. What is the impact of statements and technologies?
The general speil that is being sold now is that don’t worry about your money wasted in DW, where you got 50% or less users satisfaction, we have a solution that will now eliminate data and the warehouse and provide you with science fictional truth, every single time. Probably then it may take you another 10 years to benchmark as the non-DW solution is just starting and the DW solutions already have global stats available.
Recommendation: The solutions are there. The menthods are simple. Define the path and standards and the goal is achievable.
1. No DW: We have a new solution. Only it will cost a little more. But look at what all it can do. Caution: Have we heard that before somewhere. Recommenation: See what works and what will bring you best business value. Talk to customers, do a SWAT analysis if in doubt bet an indenepent business valeu architect.
2. Architecture is not important: I’m not sure where that goes. If you want instant real-time check on key stuff then maybe. But most enterprises do not live on real-time checks. CxO’s perform complex comparisons, trends, forecasts, cross application analytics. Try that in real-time db access mode, it is a rude awakening. Caution: Review with a BI Value Architect before finalizing a decision. A few thousand dollars could save you a few million. Recommendation: If in doubt ask an expert. Make sure the expert is on your side of the pond.
3. Modeling is meaningless: Possibly the most important aspect of DW, here I am on the side of Ralf Kimball and all his theories, except one the 70:30 ratio. Use automation modeling tools and turn the pyramid upside-down. Caution: Modeling is great in every situation. Send an email of your systems and I’ll send you a reason or two why. Recommendation: modeling impacts your information quality, object footprint, query performance and data load performance. Improve all with automation.
4. No databases: In these exceptional examples data elements never become information elements, i.e. data is not required to be stored. Caution: Great for real-time analytics, exceptional for alert reporting but not for CxO analytics. What about historical comparisons, data inconsistencies between disparate systems and most important of all performance which is the critical goal of 2009-13 BI. Recommendation: Think DW and review all else.
5. No Waiting: Also on the table are solutions that will not take 30 seconds or even 3 seconds. Ask a question and the answer shall be provided. VRR, or very rapid response systems are now coming online. Caution: What about accuracy of information. If we could not deliver a good answer in 30 minutes will this system provide it in one. What about historical comparisons, what about trends. Recommendation: Were promised this in the mid nineties and will be promised this in the mid fifteens. Check for business value and get them to sing on the dotted line the same as you do to make good the payments.
6. Throw money have solution: The general value analysis is if it costs more it must be good, so the pitch is we have a great solution but it will cost you a little more, value was never cheap and don’t we all know that. Caution: Larger spend has no impact on success of BI projects so far. The writing on the wall states that 50% of implementation, or 50% of the queries delivered, or 50% of something that has basically failed. So by the averages if you invest a hundred thousand dollars you stand to loose fifty, of one million then five hundred thousand. Recommendation: Unless you plan it well more money may just turn into more wastage.
7. Value is not attainable: What we provide is assured to provide high value to customers. We assure you 100% satisfaction, market average is only 50%. Caution: What you end with is lower than the average. Value, like best practice, is subjective. The question is whose value. If value is that the big guys delivered then why is the reported success factor languishing at 50%. Are some of the failures scared to admit failures? Recommendation: Define value and business value metrics and find the delta. Very interesting results for CFO and CxO decisions.
8. Getting Business Value/Satisfaction is an enigma: We have stats to prove that if you get 60% satisfaction you are in good hands. Caution: Aim for the moon you land on the tree top, aim for the tree top you land on the table, aim for the table you fall on your behind. Recommendation: Don’t let anyone let you accept any solution that delivers less than 80% assured. Not by quantity but by quality attach some controls and definitions of what and how it will be measured 3-4 weeks after go live.