Information is the oil of the 21st century. We generate enormous quantities of data to improve business decisions day by day, and there will be more still in future thanks to new technologies: the Internet of Things, cloud services, streaming and big data, among others.
Business Intelligence (BI) projects are the first layer of analytics — the descriptive one. They focus on automating day-to-day decisions on the basis of predetermined management concepts and data models that deliver information live. Best practice for implementing analytics projects generally recommends starting with BI before jumping to prediction and machine learning (AI), and the reason is simple: you have to walk before you can run. Management models rest on structured data models (in principle), and once those are populated they can be used to predict — and, as long as the flow of data is constant, to learn.
For more detail on this topic you can read the article: Business Intelligence or Big Data?
The method for automating our decisions runs through orienting the organisation towards analysis. Because organisational roles are similar, they adopt dashboards for managers and information cubes for analysts in their processes (see: Applied analytics: what is a management system?). All of this passes through a stage of cultural change: processes have to feed decisions with live data, bearing in mind the two BI trends — corporate and self-service.
The return on investment (ROI) of a BI project rests on all of the above, and its success will be measured in greater profitability for the company: more revenue, lower cost, more efficiency. The following points are vital to that:
1. A policy decision: there should be no bottom-up changes; every change comes top-down. The project needs a strong sponsor with clear guidelines and decision-making authority.
2. Methodology:
We recommend reading the article BI project management: best practices.
3. Model the management concepts around each area's decisions rather than around isolated reports. The result is dashboards that automate decisions. The optimal model is corporate BI combined with self-service BI.
4. Technology: the technology decision is simpler today, with a range of BI tools built to excellent engineering standards.
5. Democratising information: the aim is to eliminate islands of information and stop professionals hoarding data. Data synergy compounds; the effect is an incentive to creativity, with decisions supported by real, live data. Data becomes information after a process of analysis, and without that it is impossible to generate insights.
6. Automated, cross-cutting transactional processes: every area of the company has to take part in data traceability and in following the processes that feed the information.
7. A single version of the truth: everyone should see the same numbers; there is no room for interpretation.
In our experience, applying these seven points generates return on investment through the following (AYNITECH survey):
Operational efficiency: 65%
A single version of the truth — customer, product, supplier: 54%
Competitive advantage: 41%
Better management access to corporate data: 38%
Increased profitability: 35%
Support for digital transformation: 33%
Improvement to production processes: 20%
Improvement to customer service processes: 15%
Automation of data-related processes: 10%
Other: 4%
Nucleus Research measures the ROI impact by analytical maturity stage as follows:
Initial stage: [i] average annual ROI 188%; [ii] impact: report automation.
Tactical stage: [i] average annual ROI 389%; [ii] impact: analytics to improve decision-making processes.
Strategic stage: [i] average annual ROI 968%; [ii] impact: analytics aligned across the organisation and strategically.
Predictive stage: [i] average annual ROI 1209%; [ii] impact: analytics applied to predictive models and to social networks.
