Does self-service BI need self-service hierarchy management?
Updated: Jun 13, 2018
The business intelligence market has moved towards a self-service approach where the business users have control over their reports and dashboards. Vendors such as Tableau, Qlik, etc have shaken up the BI market and gained share by putting the business user at the center of business analytics activity.
Many voices are raising concern about this self-service approach. Is it sufficient to have access to an easy-to-use tool for producing meaningful, accurate reports and analytics?
Unfortunately, great visualization tools are only successful if they rely on good data. Thinking broadly, "good data" include dimensions and hierarchies, which define how your business is (or could be) structured and organized. As Mr. Nicely points out: "The true value of analytics is in the aggregation of data into a meaningful and useful structure for analysis". Those structures are hierarchies (and all the alternate hierarchies your business teams can cook up).
We think that before a self-service BI tool can create value, business teams need to agree on a few things:
Conformed dimensions, that define key master data such as customers, products, market segments, financial accounts or regions. Dimensions need to be unique and consistent across all analytics (you don't want to compare apples with oranges),
Standard and alternate hierarchies, in order to define ownership structures, rollups and consolidations, or classifications of things and parties.
While this is usually done in the data warehouse with IT teams that have database and ETL skills, the question is: Can the self-service approach be applied to the management of dimensions and hierarchies?
How might this play out in the real world? Imagine you're a consumer products company. In order to increase market share in the children's diapers marketyou've acquired a line of luxury/organic diapers (and consumables, e.g. wipes, creams). However, within your organization you already have an organic line of your standard diapers. The question is where should this new line go?
In this case, brand management would develop several ‘what-if' scenarios:
The luxury and standard/organic options become standalone lines,
The existing organic line is reverse merged into the luxury line and becomes the tier-2 option, or
The existing standard/organic line is eliminated to force existing customers to upgrade to the new luxury/organic products.
Keep in mind, that these alternate hierarchies will have an impact on other hierarchies, such as the financial and performance accounting, etc.
All these variations of the future dispensation of the new luxury line of diapers are alternate hierarchies. It's plain to see that these classification and reporting structures must be in place before any analysis can take place; and it's plain to see that business users are the ones that are best suited to create these alternate hierarchies.
We think that giving business teams the capability to design and maintain hierarchies and dimensions is necessary for a true self-service approach to business analytics.