By Ibrahim Shafiu, Business Intelligence Services Coordinator, Waikato District Health Board
Self-Serving in the Business Intelligence (BI) space has many definitions within the Industry and expectations vary from organization to organization. Self-serving is defined as “BI Tasks the business users carry out themselves instead of involving the IT” (BI-Survey.com), and Self-Service BI Tools ” are defined as “tools which allow users to filter, sort, analyse and visualize data without involving the organization’s BI and IT Teams”(what-if.com). The desired outcomes dictated in these definitions are an ongoing battle between the IT department and the rest of the business in many organizations.
The accessibility of various sets of data to business users has always been a challenge within organizations. The IT infrastructure and the underlying applications, which collect and store data come mostly under the IT department. Given the logical distribution of various data sources and its various business owners, it becomes a challenge for data analysts to converge all the business-critical data into one single data source which can tell the story of the business. Some of the reasons are:
1. Data Silos: A set of users, for instance the finance department, does not trust the rest of the users with their data, in terms of security and confidentiality (trust).
2. Transactional Databases (Absence of Reporting Data Stores) Managers are sceptical of people accessing data directly out of production databases, as it may affect the performance, or worse bring down the system during business-critical hours.
3. Knowledge Gap: Business users within the organization are not aware of the varying sets of data being generated within the organization and in cases where there is awareness; there is the knowledge gap in what these datasets represent.
There are many other reasons for not having a single data source, however, in my experience, the above three are the top most reasons.
These challenges cannot be solved by the mere implementation or the adoption of a BI Tool, contrary to the sales pitches from BI Tool vendors. There is no doubt that the BI Tools have the ability to connect to multiple data sources and do all sorts of filtering, slicing and dicing. However, as mentioned before, the business owners or the managers of the data sources may not be keen to give access due to trust issues and possible performance issues. In the case where there is unlimited access to these datasets, the danger is in the misinterpretation of data when it comes to mashing these various datasets. Misinterpretation in the sense on what is aggregable and what the measures represent in the business domain. Once again, the BI Tool is not the answer.
Implementation of BI and Business Analytics towards self-service is a journey which involves organizations’ top level commitment and involvement of organization wide business users . At times, this leads to the infamous data warehouse band wagon as recommended by a consultant. I am not anti-data warehouse. However, many organizations have jumped on to the band wagons of must have data warehouses to be part of the contemporary decision making. However, to the dismay of many, by the time the data warehouse is delivered, the business questions that it is supposed to answer is either outdated or not relevant. In worst cases, the business model has changed. Nevertheless, an organization needs to have curated datasets or data stores to implement a BI and analytics program. Curation is where data silos are opened, and the owners provide the meaning and definition of the datasets they generate or hold. This does not mean pre-aggregation but the datasets made available can be easily interpreted in business terms in relevance to the business domain. The next stage is the competency building stage. This is the stage where the business users across the organization are exposed to the data driven environment. There should be focus groups, which involve business users across the organization that promote application and utility of the available datasets to gain insights and to tell business stories. The organization should have data catalogues and data dictionaries published to enterprise portals, easily accessible to business users. This fosters a culture of data savvy business users.
There is a logical flow in becoming a self-serving organization in the BI Space. A powerful BI Tool is of course a must have, but as discussed, there are many parts to the puzzle that has to fall in place before investing in the adoption of an enterprise wide BI Tool. This is once again contrary to the BI Tool vendors philosophy, which infers that anyone having access to a BI Tool magically becomes an analyst and can start mashing datasets and perform slicing and dicing that can provide all the answers and insights the business may need. This is a competency that has to be built, and managers and the key decision makers have to go through a structured learning program in order for them to become the consumers of the self-serving environment. There are no short cuts to this process; however, I agree that this process should not take a long time since key business users are already domain experts and they know their figures and use them to make their business decisions. The gap to be filled is to enable these business users to put the pieces together themselves, whenever they want and however they want, without depending on others.
Professor Rajeev Sharma from the Waikato University, in his public lecture on business analytics, discussed his findings on his research to understanding why analytics do not always live up to its full potential. He emphasised the fact that to make analytics work, you need talented, experienced staff that demonstrate high level of technical skills and have the competency to evaluate risk and make good business decisions. My take on these bold statements derived from his research findings are that people using data for their strategic decisions should raise their technical skill levels. Too many a times, we see that BI Tool vendors or BI consultants developing customised dashboards for the decision makers. This beats the outcome we seek as an organization to become self-serving. One wrong drag and drop or one wrong slice and dice maybe like taking a wrong turn into a dense forest without a GPS. The user ends up talking to IS support or the contracted vendor to fix the dashboard. This defies the very definition of self-serving that is presented at the beginning of this article.
In summary, the journey towards the self-serving analytics should start with a sound business analysis and a data audit process. This will assist in identifying the key business indicators, , if the needed data is being generated within the organization and by whom, and potential gaps in data sources needed for these indicators. Once this is achieved, the second stage would be identifying the location for these sourced data sets in a reporting environment, call it a centralised data repository, data warehouse or even a data lake for that matter. These sourced datasets should have data governance structure fitted around it alongside with data catalogues, data dictionaries and use cases. Last but not the least, is raising the competency level of the data driven decision makers in consuming these datasets using BI Tools. At the minimum, an organization has to go through these stages and should have the said entities in place before an organization becomes self-serving.