Enterprise Self Service BI
From a Single Version of Truth to a Network of Truth
Single version of the truth
The primary mission of business intelligence has been to help organizations run better by connecting people to the information they need to make better decisions.
Prior to formal BI systems, only a small set of highly skilled people could access enterprise data, and the technical methods they used (writing native code and SQL) didn’t scale across the organization. So, the first step in providing trusted data across the organization was to “manage” the core data assets and expose them in a more user-friendly way. The term “single version of the truth” was coined to represent the management of core data assets so that they could be made available in a scalable, repeatable and trusted manner – usually via production reports.
Data warehouses, marts and BI platforms have been widely deployed with the goal of providing a single version of the truth. IT has been the driving force in the deployment and management of these systems and the transformation of companies to use data as strategic assets.
Success leads to new challenges
As BI gained popularity, users began asking for more information. Since data needed to be managed by IT, new information requests ended up in a queue of IT tasks. Ad-hoc reporting solutions appeared to provide end-users with a wider sandbox of information to work with – giving them a level of autonomy much greater than traditional operational reporting. Ad-hoc reporting reduced the pressure on IT and provided a release-valve to the pressure. Well, for 10 years anyways …
While ad-hoc reporting bridged the gap between IT and users for a number of years, and led to enormous growth of BI adoption, we are now seeing a new shift in user expectations and big data that has put renewed pressure on the information supply chain. Users want even more agility to use information to innovate in their local business areas, while the growth of mobile devices and big data have put a new level of pressure on IT to manage ever growing assets.
This time, it is clear that we need a different approach to accomplish our mission to connect people and data.
We need to shift mindset from a single version of truth to a Network of Truth.
The Network of Truth
We believe that new user expectations and big data require a different approach, and that the right solution is an information system where both IT and end users can contribute to the information assets – resulting in a network of information that grows in value with each user and each interaction. We call this system the Network of Truth.
The Network of Truth considers each user as providing unique contribution value to the system. Instead of IT doing all the work, and end users eating what they are fed, the Network of Truth creates a model where the work that IT has done is enriched and extended by the user community. A simple analogy to this would be the world of encyclopedias – where:
- Single version of truth ~= Printed Encyclopedias
- Network of truth ~= Wikipedia
With the Network of Truth, each user contributes meaningful value to the enterprise information network:
- IT provides the foundation by loading data, defining physical relationships, master data management, user management and governance.
- Analysts can then enrich the data by adding in local variations specific to their business. This could be anything from renaming columns, creating custom groups, relationships, calculations and hierarchies to mashing in data that isn’t yet in the network.
- End users (Decision Makers) will typically contribute opinion. They are the eyeballs that help the system understand which data is important, whether there are human relationships between the data that isn’t captured in technical relationships, and suggest contextually relevant information.
The Network of Truth is unique in that it elevates all these users to having a contribution to the overall value of the system. Other approaches that focus on just one or two of these user types are inherently partial solutions.
Now is the time for the Network of Truth
The natural question now could be, “why didn’t we do this earlier?”. As it turns out, there are a number of preconditions that are necessary to successfully build the Network of Truth.
First, to achieve a network effect, you need a critical mass of users craving enterprise information – with the success of BI over the past 20 years and the growth of mobile users, we believe the potential BI population is here. Importantly, you also need to design the network with a focus on the intentions of these three core user types so each can contribute their unique value to the network.
Second, you need a real-time data platform that shrinks down the supply chain of information and makes operational data available for analytics in real-time. With SAP HANA, we can do analytics on transactional data in real-time – without the need for IT to have multiple intermediate steps of data extraction and preparation. This allows IT to provide real-time analytics to a big user population at a lower cost of ownership. It also allows analysts to enrich the data with their own views and calculations without the system incurring additional materialization costs.
Finally, you need a system that can provide “speed of thought” answers to user questions – not just in small silos of information, but across the information assets of the company. Again, with SAP HANA, we have the ability to quickly search across the Network of Truth and return meaningful answers to business users – which is the ultimate goal of an ever-improving business intelligence system.
Applying the Network of Truth
Let’s have a look at a business scenario to illustrate the value of the Network of Truth in action.
Consider a marketing team that wants to run focused campaigns based on brand sentiment. They also need to ensure that the budget allocation reflects product growth expectations. In this case, IT provides some of the baseline information, such as existing product performance. The other two sets of information come from other sources: the plan information is contained in spreadsheets, and the brand sentiment information is available via a web services subscription.
With the Network of Truth, the analyst in the marketing team can mashup the corporate product performance data with the plan and brand sentiment data. She can fix errors in the data, create formulas, and groups to adjust the data to her local context. When all of this is done, she can then save her work back into the Network of Truth. This means that the new relationships and enrichment she has done is now stitched back into the corporate information assets. IT can then decide whether they want to apply the changes globally, or simply leave them as local definitions for the marketing team. The marketing team can run new scenarios quickly since previous work is available in the system, and they potentially have access to data and metadata contributed by other teams across the business.
The evolution of the metadata, data and usage of the Network of Truth make it easier and faster to answer new questions, improves the quality of the data and, importantly, provides a broad view of the signals affecting your business.
Evolution of the Network of Truth. IT provides data management, analysts contextualize the data, and end users provide usage-driven metadata.
I wonder how we will define 'data quality' in this new context.
Data Quality has traditionally been a one-way street with IT shouldering all of the burden of ensuring quality. Of course, Analysts have always taken IT data and enriched it (including performing quality improvements), but all-too-often this contribution by Analysts hasn't made its way back into the core data assets.
The Network of Truth is really about having all users contribute to the data assets (in their own unique ways). In the case of data quality, one goal of the Network of Truth is to provide a way for Analysts to save their enrichment back to the Network. In itself, this should be a big improvement.
Over time, we'd also want to have the end-user play a role in data quality by being able to rate and comment on data.
Ultimately, data quality should improve with the Network as more users (and different types of users) all play a role in incrementally improving the quality of the data they care the most about...
Thanks again for your feedback!