BI Strategy & Definitions
In the previous Blog I spoke about the need for an effective Enterprise Intelligence strategy. In this blog I want to delve deeper into what a BI strategy is and the key models that make up a successful system.
When looking at the overarching strategy, people usually start with KPIs (Key Process Indicators / Key Performance Indicators) that show how an organization is performing. Often, these KPIs will be related to financials, efficiency and/or yield. It depends on the particular target area you are putting your Business Intelligence in and it would also encompass any definitions. For instance, some people calculate yield differently and therefore it would use the target area’s definition for yield calculations. Some want global yield, some want process yield and some want to know overall yield through multiple operations. They want to accumulate this knowledge to understand min/max lead times for reasons such as informing their customers of the time difference from ordering to receiving a product or how quickly do they need to be responsive to the change in market. This strategy and solution do not have to be solely focused on manufacturing processes; It could encompass many business functions such as quality release, order processing, planning, supply logistics, etcetera.
What is also important here is to take an inventory of your existing systems. A lot of organizations already have some systems collecting this data. A company wants to leverage those systems as much as possible (if they are desirable and serving their purpose adequately). An organization must identify where those systems are located and what they are recording in order to understand what data is being collected and potentially how much redundancy is there.
Due to the proprietary nature of these disparate systems, it is also important to develop Canonical Definitions to clear up any discrepancies across business units, systems, etc. so that every system is speaking the same language at the Business Intelligence level. In other words, harmonization of languages across your applications is essential. For instance, the details in a Production Order must be understandable by everyone in the organization at all levels.
The Canonical Definitions are to be used in your BI Models to insure proper contextualization of data and that each application is providing the appropriate level of detail in order to ascertain the appropriate context for the data provided.
Defining an action plan is critical to insure clear expectations are set for the use of the data. Some organizations try to implement a data warehouse and reporting tool hoping it will somehow catch on and people will adopt it and just start using it. However, if you do not define the actions initially, how do you know the right data and tool was selected or if it is even adding value? Once the “new” wears off, it loses enthusiasm and is just another application for your IT group to support. By defining an action plan, not only are expectations set, but requirements for the solution to insure it meets the overall objectives of the BI Strategy are also built. This can be important especially throughout the vendor selection and project post-mortem assessments as the solution can be measured against these core objectives and anticipated objectives for the target area.
Even though the Enterprise Intelligence Strategy is defined in a holistic manner, it is important to constrain the scope for each phase to enable you to roll out the BI solution and develop a very clear action plan in a controlled manner without losing sight of the objectives.The strategy should include a Scope and Phase definition in addition to the applicable action plan for that given phase. Begin with a specific target area in an organization to start the rollout, then list subsequent rollouts, leaving room for improvement after each implementation to allow for continuous improvement and expansion of the solution as the scope expands. This is ideally executed in an iterative cycle – PDCA approach (which is a lean methodology for Plan, Do, Check and Act).
It is ideal to start in a specified target area. Plan what you are going to do. Do it (execute the plan). Check and evaluate the results at the end of the phase. Make adjustments and Act appropriately to improve when you move on with the next iteration or phase. Then based on what was learned and experienced, start a new iteration with a refined plan. Therefore, organization’s solutions are improving in addition to their associated processes, through each iteration.
If a company attempts to do everything at once as a “big bang” approach, it will more than likely fail. The processes will begin to adapt and improve requiring the solution to evolve as well. We have witnessed greatest results when a company allows the solution and the process to evolve together.
A typical action plan iteration will address the following:
- What is to be measured, used, and communicated
- What will be open for discovery models
- Alert & Notifications scenario definitions
- Business Rules for automated responses and actions based on certain KPI violations.
When evaluating the Business Intelligence models to create and use, it is imperative to understand the intended use and type of data used in order to accurately define the capabilities needed from these models. The best way to understand this is to take a look at a few example models and understand how they can relate to specific initiatives and needs.
- A KPI model
- A Static Aggregation
- A Discovery Model
- A Predictive Analytics Model
Check out the next Blog to see how Applications play a critical role in an Enterprise Intelligence Solution and can greatly assist with any continuous improvement activities.