Why Doesn't BI Catch On?
-
I've heard for many years that each year is the year for BI. It's finally catching on, it's growing, more companies are implementing it, etc. Yet I still don't know many people that are really using Analysis Services very much, we get little traffic on the site from BI people, few questions, and it just doesn't seem to be "mainstream".
Then I ran across this article that talks about some of the BI stumbling blocks that companies face. Yet it doesn't, I saw one quote about non-standard APIs being a problem, but most of the article still looked at the fantastic ways in which BI systems can help people. There was a mention about security not being a concern with sensitive data being analyzed, but it's never a concern until something happens.
Now I know lots of you BI guys will say BI is great, you're working on things, it's being deployed, but the reality is that it's a far, far cry from mainstream. MDX is little used, even by relative standards, it's little used. The front end tools are sold infrequently, and lots of companies (small to medium) or departments in large companies, aren't tackling projects.
As I see it, there's two reasons for this. First is the cost. The per seat cost of most front end tools, Cognos, BO, etc., is high, very high and unless you have some specific project with definite payoff, it's just not worth "experimenting" with BI. Especially now that reporting services gives you a quick and easy reporting tool. Even some of the low end tools I've seen at SQLServerCentral.com are a few thousand dollars per seat, which for any size project is pricey. Most sales are still six or seven figures, which is out of most budgets.
The other thing is that moving to dimensions, measures, MDX, etc. is a new paradigm and most DBAs just are not comfortable in these areas. Most of us are still working on learning more about relational systems and so any type of project that involves Analysis Services now costs consulting dollars as well. And it means that the primary DBA is no longer the primary DBA for this system.
Don't get me wrong, I think BI has some great uses. And it can be incredibly valuable to a business. It's just expensive and a niche type technology for some time. Much like many other technologies that are slow to catch on and unlike some technologies like "VoIP" that caught on quickly, BI isn't catching on anytime soon.
Steve Jones
-
The TechEd on SQL2005 made a big play about BI.
The general gist was that Microsoft had been surprised by the uses that people have put DTS to with regard to data-warehousing and analysis.
With BO increasing the price of Crystal Reports Microsoft's share of the BI market has shot through the roof. For this reasons DTS, analysis services, reporting services, notification services have been integrated into SQL2005 rather than kept as bolt-ons.
I used BI 5 years ago with one of the brands of a major European catalogue company. The results of one particular analysis were astonishing and without BI the insight gained would not have been available. I think BI is a sleeping giant that is beginning to awaken.
-
I've implemented Cognos BI for IM/IT operational planning / reporting in two of Canada's biggest IT shops. While both applications are enormously successful they resulted in an interesting lesson learned.
BI is a very powerful analysis tool and is typically sold to senior management in a demo that quickly presents ad hoc analyses of large quantities of enterprise data in visually appealing formats. The sales pitch is usually about empowering managers with this wealth of information that they can instantly access and analyse without programming.
There are the usual stumbling blocks of training and getting the data in different feeder systems to agree but the real uptake issue is often one that the sales rep won't tell you about, as he tries to sell you BI licenses for everyone in the organization - most users don't need enterprise BI.
The power to do computerized ad hoc analyses of enterprise-wide data is of little use to middle managers whose scope is local. Yes, they can benefit from desktop alerts and dashboard technology but their followup analysis usually consists of picking up the phone and asking questions directly. Turning data cubes on their head is a waste of time when you know that one phone call will not only provide an in-depth analysis of the issue from the source but also enable you to get it fixed. Good managers tend to be people oriented and would rather talk to someone about the issue than analyse it on a computer. It's a cultural issue, not a technology issue.
Now don't get me wrong. BI tools are without equal for some applications and some users (besides they are very cool). Business analysts who must support senior management or enterprise wide programs with ad hoc analyses can benefit enormously from BI tools such as the Cognos toolset. Since they are supporting senior level decisions the value of the analysis can often be highly leveraged. However, how many of these analysts are there in your organization?
In one of my implementations, only 15 out of 300 users qualified. The others wanted dashboard type notification of issues within their mandate ONLY. Often their concern was not an ability to analyse the issue - they already knew more than they wanted about it. Many managers were more concerned about what was being reported upward about their shop, in order to be ready for the inevitable questions.
Yes, BI tools can meet these dashboard needs but they carry a high price tag. Often there are cheaper technologies to provide "red light / green light" notifications to the user community. Unless there is a real need for the individual user to do cube type data analysis on a regular basis, the money spent on that BI license might better be spent elsewhere.
Ric Kersey
-
I completely argee most users have no need for BI and most administrators/managers want such specific data when scrutinizing something that the cube is of little use. For the cost of most BI solutions, companies would be better served to just transfer those $$ to data analysist positions and just make a data request go through those people instead of making a manager try to figure out Cognos, BRIO, Corporate Radar etc.
Most BI tools (such as Cognos) are way too complex and cumbersome for all but IT staff anyway. I have been warehousing since 1998 and I have yet to see any ROI. I have also implemented very cheap solutions such as the OLAP thin web client from the resource kit. This was a extremely easy/cheap solution with analysis services and still when people really need important reports they would not use it and I ended up doing the report for them anyway.
My opinion is to keep the warehouse for its capabilities to hold/aggregate your companies data, but ditch any ideas of having any user be able to run any report anytime they desire. And skip any money that you would pour into a OLAP front end, they are simply too complex for end users and too dumb for IT staff (try to develop a simple query in Impromptu, its all click and drag you can type no SQL!). Spend the money on people not the worthless software currently out there. If you must do something for your managers look at dashboards and call it good.
-
Business Object 11 can be used by an executive but SQL Server Analysis Service require development skills. BI have not been a boon because relational calculus is not what most SQL developers use most use relational Algebra. The other problem is businesses being cheap I was once in a meeting, the company's first BI project and everything was in relational Algebra and the project was close to deployment. So I say cost and skills but the upside for SQL Server, I sent the UDM(unified dimension modeling) link to my SQL Server SIG list and one of the members said he is doing something similar in Hyperion so SQL Server 2005 will change the market.
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnsql90/html/IntrotoUDM.asp
Kind regards,
Gift Peddie
Kind regards,
Gift Peddie -
My experience with BI is that it tends to be targetted at the Sales/Marketing functions.
I have yet to meet a sales/marketing person who is au fait with even the basic maths behind the system. When confronted by an observed trend they tend to either dismiss it as a glitch or explain it away in some other way. By the time they see that there IS a trend any opportunity that BI could have given has been missed.
-
I have worked in BI for the last 5 years now and I have to admit, there has not been a lot of change in terms of new functionality. Everything seems to be rehashed to work on the web rather than a full client tool.
Using MDX and OLAP is a good idea but is notoriously hard to firefight in an emergency, whereas a warehouse populated by stored procedures makes it easier. I tend to find that a lot of managers do not trust OLAP. Unless they can write a query to check the data they are not willing to invest. Personally, I would use OLAP 90% of the time if given the choice. The speed is incredible when running queries. It just needs a good front end that is easy to use to make it more mainstream.
I also agree that a lot of staff cannot get to grips with the way these systems work. But there are a lot of hands on middle management gurus who do make good use of the tools available.
As for the issue of ROI, I think it depends what type of business you are in. Additionally, there must be objective set every time a report is created.
For instance, I work in insurance. One of the key objectives is to review a list of risks due to expire in 3 months so that underwriters can make plans to renegotiate a contract with the broker/s concerned. Having a scheduled monthly report sent to the necessary parties is paramount and could result in loss of business if not in place.
This brings me to the point where I feel BI is not used properly. Allowing external users into your warehouse can offer so many benefits. The above scenario would be great for brokers to look at their own risks within my company and perhaps the insured themselves. Adding mail links etc would also greatly enhance the contact between all the parties.
However, this is where we get into the security issues. Infrastructure guys don't want external users coming in. Managers are scared the information will fall into the wrong hands. The list goes on. A proper security implementation would safeguard everyone's interest.
Anyway, enough dreaming for now. I'm looking forward to X2 which is a merger of business objects and crystal reports. I just hope there's a better front end and nothing too complicated for an average point and click user.
------------------------------
The Users are always right - when I'm not wrong!
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply