September 24, 2018
As organisations become digitised, we find ourselves managing collections of systems that reflect individual parts of the total business model.
Whilst the ability to use specialised applications to manage core activities has created huge benefits, it’s led to another problem: we’ve created a world of information silos. A business is a complete end-to-end process and not just a collection of functional sub-sets.
Information silos can be likened to a physical storage silo, typically used for grain with the same characteristics by design:
That works great – for grain!
Consider the following collection of database systems, from Marketing Automation through to Financial Reporting:
Look familiar? They’re all individual silos of information that currently exist within many organisations and, although there may be some integration between them, rarely is all of the data contained accessible as a collective whole.
The problem is often further amplified with sub-silos within an individual system. For example, a Human Resource Management (HRM) tool may be able to report on time and attendance and may also contain information on pay scales, but it’s unlikely to be able to put both of them on the same chart and lead to a deeper insights, like how pay affects time and attendance. The good money says you’d have to export two or more sets of data into Excel and merge them back together to get that kind of data visualisation working.
This is a sensible setup for grain where you may need to reduce risk of cross-contamination, or you may want to store different grains and/or harvests separately.
However, this is not so ideal for data. An organisation is a collection of actions and things those actions apply to. These ‘things’ don’t respect the boundaries of the separate systems. The same staff member who fills out a service request in the Enterprise Resource Planning (ERP) tool may also have a sales target in the Customer Relationship Management (CRM) platform and an expense claim in the payroll system.
In fact, look at the order I put the database silos in – they aren’t just dealing with overlapping business entities, they are a set of links in a chain:
Issue 1: Siloed reporting does not allow you to easily see ‘disconnected’ data in context together. For example, take an organisation with 5 metrics that it uses for health indicators: lead conversion rate, sales dollars, customer satisfaction scores, resource utilisation and staff turnover. Traditionally, the only place you will find all of these measures together is in a manually collated report and even if these scores are presented together in a single dashboard. They cannot be sliced together, so you cannot simply select a sales rep, region, customer industry or resource type to slice through all measures at once.
Issue 2: Information silos tend to lock us into a simple pattern of analysis, comparing only points of data that are contained within a single system or sub-system. We can be so accustomed to this approach that it may not even occur to us to ask questions that cross these artificial boundaries.
The solution is to integrate the relevant sources of data into a single data-mart.
The above diagram is a standard approach to data warehousing. The data mart (star symbol) is the central source of truth in the reporting system. Merging data and structuring it for analytics is the key to integrating your data.
Not only does this facilitate data being analysed in context, but it also empowers asking the more holistic questions. Moreover, it allows you to do it personally, regardless of your technical background.
The goal of self-service Business Intelligence technology is to allow the right people to ask the right questions as needed, regardless of what systems the data originated from. Achieving this depends on designing a well-structured data mart – and that’s my job.