[OBIEE 10g/11g] Best Practices

BI Best Practices

Often some people like the coding best practices and then once again the architectural best practices. Often in BI world, the best practices are driven from the business side. You understand what I mean.. The more you know how the business works, the better are going to be the dashboards, the RGY charts and KPIs. None of these makes sense unless you convince the business to use them. How do we make the business folks use them? This is possible only when the requirements come from true business.

When coming to the dashboards, they can be categorized as many ways as you like. This is something that I found works in real world (atleast based on my experience).

 

1. Executive Dashboards

These dashboards may remain constant for the long term except that there may be changes in goals specification. When these dashboards are designed, all the KPIs are supposed to be long-term strategic objectives that any organization’s executive management would monitor. Never add short-term KPIs as part of the executive dashboards. A rule of thumb that I use is that these KPIs should have existed in the organization atleast for the last one year and the buyin from executive management that these remain strategic KPIs for the next 1 to 2 years.

Most often, the executive dashboards are refreshed nightly or weekly based on the type of metric we monitor. Daily monitoring metrics i.e. sales since last night, Change in AP (Account Payables) by department etc. should not end up here in executive dashboards.

2. Operational Dashboards

These dashboards are for middle level management i.e. department managers and below to monitor their every day changing KPIs. Ex: In Manufacturing domain, we try to monitor the number of parts we manufactured in that shift, number of parts we scrapped, what is the turn over ratio etc. For sales domain, lets say, how many did we sell since last night?. Other metrics that makes sense for each department need to be in their departmental dashboards. Who can help in underdstanding what goes into the operational dashboards? The quick answer is Mid Level Managers who monitor their departments day after day and who look for some numbers every day morning on their desk. I had once checked with one of my clients who had 2 of his department personnel start work early in the morning like 5 AM so that these reports are run manually and be ready by 8 AM. Atleast now, because of some improvements made in the department, these people have these reports ready by 7:30 AM and they take a look at them just before 8 AM presentation.

Specific to BI projects, these are some rules of thumb I would always make sure we follow.

1. Involve your business users as part of the BI project from the initialization phase.

I have seen both sides of the coin wrt this. One project where the IT has come back and said we know what our users use. We have been monitoring the reports usage for the last 6 months. We will take the top 50% reports or top 50 reports based on usage and create them in Phase I. We will let business folks come to our meetings for phase II and let them ask for modifications of Phase I reports and other reports they would like to be migrated or upgraded.

This project failed in getting the buyin from the business folks and infact the budget was cut for Phase II because there was no input/ negative feedback from the customer side.

2. Never try to replicate whats in there before while upgarding or migrating.

When you migrate or upgrade your BI software, never try to replicate whats in there before. Always, go back to the customer and ask them how they would like the reports/ dashboards to be changed. Couple of changes here and there might actually delight your customer.Also, lets say when you are migrating from one BI tool to another, know the capabilities of each tool and try to advertise the new features so that the business folks might understand and make better decisions on how to use the BI reports and dashboards.

3. Dont expose all features of your new BI tool to your business.

This, one way, is cheating your customers. But, some times it works. There are times where the carrot and stick approach does not work nor the customers does not give you any feedback and finally the word comes out from the business “Replicate whats in there right now. If this new BI tool does what the old tool is capable of, I am good. I do all my analysis in Excel. There is no way we can create standardized business rules to my data analysis”. I had customers where all they wanted is excel based reports sent out to them from a scheduler every day. 

One of my IT managers, actually cheated the business folks saying this new BI tool does not have real nice capabilities to send excel reports based on a schedule. Did the business buyin with this? No. But, they atleast rolled up their sleeves and started documenting what they do with all the excel based reports. Now this manager showed these folks some dashboards where all the manual steps that the business folks were doing is already incorporated in the dashboards. Finally, he showed them how to downalod these dashboards for presentations, exporting to Excel or what ever. What  were the benefits of cheating the business folks initially? 

This made the business folks communicate to IT what they really do with the data. Other reason being, the new tool’s capability of creating dashboards and the old tool does not.

4. Easy to use.

Dont try to complicate the business folks with how to find data. This is the last thing you would like to go through as an IT person. Customers trying to call you where to find data and what each color in the dashboard means.

Make data available to use as easy as you can. A recent analogy I found interesting was ebay.com. This company has never helped users on how to find each item on their web site. All they have is a “Search” text box on top of their web pages so that all users can type what they are looking for.  The other feature that I found useful is “Add to favorites” or “Shortcuts”. Once you found what you want, you can easily go there next time without going through the same hassle that you had initially.

If you could incorporate these two features into your BI portal/ dashboard/ website, that would certainly delight your customers.

5. Dont clutter your report with every single detail on one screen

Never try to put each and every detail in one screen. Lets say there is a report regarding “Profitability By Vendor” report. Having each and every detail on one screen regarding how you came up with Profitability number on one screen would confuse the customer. Just give him the final number. If the customer really wants to find out how you arrived at that number, create a seperate screen for that like “drill down” or a link from the main screen.