BI Ops: Governance for the Consumption Layer
The consumption layer can’t be overlooked, and has become too large to leave to data governance strategies focused elsewhere. BI Ops is the answer.
The most useful tools are able to be used and understood by many people, while making their lives more efficient and effective. By that measure, business intelligence (BI) has proven to be extremely effective in the area of data and analytics. In recent years, the emergence of Microsoft Power BI, Tableau, Looker, Qlik, and a growing number of tools have put powerful reporting, analysis, and visualization capabilities into the hands of non-technical business users, in turn helping organizations unlock the power of the vast troves of data that are a crucial asset for modern enterprises.
Today, BI has emerged as the interface between an organization’s data and the business users who derive insights to present to decision makers. It is at this consumption layer where most people interact with data directly, and, increasingly, professionals are coming to expect self-sevice access to BI tools so that they can dig into available data and create reports, dashboards and visualizations. Organizations have encouraged BI adoption, especially as those insights drive the decisions that help enterprises reduce waste, optimize operations, adjust to consumer signals, and stay ahead of trends.
As a result of increasing demand, BI is a growing segment of the data and analytics market. Seeing the power of data has led organizations to collect more of it, especially as a growing number of monitors and sensors make it easier to collect data from an exploding number of apps, devices and internet-connected machines that are increasingly central to our working lives. Meanwhile, the number of available BI tools has skyrocketed in recent years, while the emergence of generative AI stands to scale their capabilities. With more users generating more reports on more platforms, BI is occupying a growing share of enterprise resources. According to McKinsey, BI and reporting make up 5-10% of total IT spend.
Yet, at many organizations, management of BI has not caught up to this growth. The explosion of data has led many organizations to invest in data governance at the warehouse level. This has helped to centralize a set of standards that centralize, manage, and secure data. Yet organizations are now finding additional strain at the consumption layer, where most of their employees access the data. In particular, there are considerable challenges associated with managing multiple platforms, which is the norm at most large enterprises. Each tool is a siloed environment, often with its own set of rules and procedures. This means organizations have conflicting standards on how to define key data, validate datasets, and govern access. This results in duplicated efforts, conflicting reports, and even presents potential security risks.
While data governance solutions focus on addressing these issues, their focus on the warehouse layer leads the challenges to go unaddressed in BI, even as the proliferation of reports with duplicated, inconsistent, unused, and freely permissioned data undercuts the strong controls in place further down the stack.
If left unaddressed, these issues could prove to become a drag on the considerable momentum that BI adoption currently enjoys. Creating two reports about the same topic that present the same metrics in different formats but show different results is not only a waste of effort, it can also create confusion among those who consume the reports. In turn, this erodes trust in the data, and could lead to doubts about BI. Given that BI is the point where data is accessed by the most users, it could also erode trust in an organization’s data program as a whole.
Just as there is a need for data governance at the warehouse layer, organizations must adopt BI ops at the consumption layer.
In this report, we will cover how data governance gave way to BI Ops, and a look at the steps necessary to implement a BI Ops program.
Why Governance Doesn’t End at the Warehouse Level
The democratization of data has created massive opportunities for teams across an organization to leverage analysis capabilities. Yet, for large enterprises, it is also the reality that managing this data is necessary to drive adoption of analysis capabilities at scale, and ensure that data is harnessed for effective and productive decision making. To promote adoption, data should be:
A Source of Truth: Can data that is from reliable sources be accessed in a streamlined environment that is organized, free of redundancy, and managed through clear processes?
Trusted: When users build reports and managers view them, do they believe the underlying data is accurate and presents the world as it is, not based on wrong information or presenting an argument that represents the presenter’s pre-conceived opinion?
Secure: Is access to data permissioned appropriately, and are safeguards in place to keep data from leaving an organization?
In order to achieve these three principles, organizations have implemented data governance programs, which are implemented to manage data in the data warehouse, where it is stored, organized, and aggregated from multiple sources. Data governance programs are focused in the following areas:
- Data Unification, bringing data from multiple sources together in one place, in one view.
- Data Cataloging, particularly the creation of a centralized inventory of data assets.
- Search and Discovery, providing users with capabilities to easily locate data that is relevant for their use
- Data Quality, ensuring that the data is consistent and de-conflicted, up-to-date, delivered from reliable sources, free of duplicates, cataloged with unique identifiers, complete, and accurate.
- Access Controls, governing who can access particular data, and establishing standards around use and exfiltration.
As more organizations implement data governance, a growing number of platforms has emerged to assist in each of these areas. Alation, Atlan, and Informatica are just a few of the vendors providing data governance capabilities at top enterprises. Increased data and the growth of cloud storage are expected to propel growth in the data governance market from $3.02 billion in 2022 to $14.53 by 2030, according to Fortune Business Insights.
The growth of data governance is a testament to the need that organizations see to implement each of the pillars of data governance noted above.
Importantly, however, data governance platforms fail to implement governance in the consumption layer. This is the point where the vast majority of people within an organization access data to create business intelligence reports and dashboards. It is also the point where the data leaves the purview of the team tasked with data management, and enters the realm of business users who lack the technical grounding and training that can reinforce sound data hygiene practices.
While organizations have made strides to manage growing volumes of data, they are still catching up to the explosion of business intelligence use that ultimately puts this data to work for the organization.
Inside many organizations, there is a belief that BI and the consumption layer is already covered under existing data governance programs. After all, an executive has taken the necessary steps to identify the importance of data governance, create a program and resource it with a team, and bring in a platform that addresses the technical requirements for success.
A whole second layer of transformation occurs as soon as data hits the business user. This challenge is compounded by the fact that most organizations have more than one BI tool. In fact, according to McKinsey, 70% of companies having three or more BI tools. Each tool is its own unique environment, meaning there are differentiated ways of managing data, meting out resources, and managing compute and billing across different platforms. At the same time, these different tools are often accessed by different teams across the organization, each of whom may have their own ways of presenting data and enumerating values. This is part of what is driving the explosion of BI reports, and these reports are being created in an environment that is not governed.
In fact, the issues that Datalogz observes in BI environments are in direct violation of the three principles of data governance outlined above, namely that data is easy to find, trusted, and secure. Let’s take a look at some of the key challenges:
Source of Truth
As noted above, different teams operate on different BI platforms, meaning they access data in different ways. This can create duplicated data, and duplicated efforts.
In one scenario, we saw how multiple teams created duplicate datasets and reports. This not only leads to conflicting insights, but it also means that data storage is being used to store the same information in multiple places. So costs go up, and the picture for the organization is no clearer for the effort.
One particularly problematic area across many organizations is a lack of a clear process for endorsing datasets or reports. Without it, teams may rely on unverified data sources. This leads to inconsistent reporting and decision making, as different groups might use varied data definitions and metrics.
In one case, an unendorsed dataset, which was not validated or monitored, was being used across 40+ reports. In turn, those reports relied on daily incremental data loads that frequently failed. Business users weren’t even aware of the issue, meaning it could’ve resulted in flawed business decisions.
Trusted
Data programs derive their legitimacy from trust that the information in the report is an accurate representation of reality. That trust has to be earned, but it can be easily lost.
Take this scenario: The finance team and the operations team contribute data to a report, but they calculate order inventory in a different way. The CFO then catches this discrepancy, and questions its validity. Ultimately, this erodes trust in the data that is being presented. Instead of using data to make a decision, the CFO is now doubting the organization’s data as a whole.
Security
While the data warehouse layer is often governed by a well-established permission structure, in the BI layer it is often the Wild West.
To be sure, business users need access to a range of data, both to complete reports and discover what is available from the organization.
But, too often, inadequate review processes fail to monitor individual user access and permissions within BI systems. After all, unauthorized access in BI could result in the mass export of sensitive company reports.
Data doesn’t have to leave the organization to cause exposure. In one case, an HR report containing sensitive compensation data was inadvertently shared across the organization. This threatened privacy, challenged morale, and left the organization legally vulnerable.
Data security is often presented as a battle between organizations and outside actors. But, in truth, the internal challenges are just as great, and BI is a primary and growing arena in which this plays out.
Why We Need BI Ops
From a high level, it may appear that the issues covered above should be covered by a data governance program. After all, they fit the mold of the priorities for data governance. The fact that organizations are still facing these challenges, indicates that a new level of solutioning is needed. After all, BI is its own environment entirely, and has particular challenges associated with it. BI needs its own set of guardrails to keep issues from dragging down data programs as a whole.
BI Ops serves as a new frame for the management of data after it leaves the warehouse. By focusing on the consumption layer, we can hone in on BI reporting, datasets, users, platform administration, and resource consumption.
Through BI Ops, we can ensure that business intelligence is a partner discipline in promoting data governance, rather than a hindrance. This is critical. Since BI is where most users interact with data, it is the point at which their perceptions of data programs and the effectiveness of management are formed.
The Datalogz Control Tower serves as an enabler for organizations seeking to implement BI Ops by providing a tool to monitor the consumption layer.
The platform establishes a BI source of truth by automatically extracting and unifying metadata. This helps organizations to understand their existing BI environments. Having visibility in an environment is the first step to rooting out challenges. When organizations have multiple platforms, they may not be aware which departments are using certain platforms, and they could have other departments employing shadow IT to perform their own BI analysis, outside the purview of the data team.
Once this visibility is in place, the Datalogz Control Tower then deploys monitors and alerting across the analytics stack.
This promotes trust in the area of governance at both the data level and the system level. At the data level, it provides statistics on asset creation, engagement, and failure, so that organizations can understand usage patterns that could ultimately present challenges. To root out discrepancies and redundancy, it also identifies overlap and duplication between models, datasets, and reports. To reduce and prevent reliance on unverified data sources, it identifies stale data and unendorsed datasets that are consumed regularly. And at the system level, Datalogz tracks changes to particular settings to catch potential problem spots before they cause issues.
The Datalogz Control Tower also acts as a bulwark for security. The platform includes tools to understand user behavior, including tracking IP addresses and geographic access, as well as dataset and report-level insights. It also monitors for irregular data exports, and asset-sharing behaviors. Further, organizations can monitor administrative and tenant setting changes for any potential problematic permissioning.
The growth of BI has made the consumption layer a critical interface between organizations and their data. To enable self-service access that many organizations target, there is a need to implement a new level of governance that goes beyond the warehouse layer. The consumption layer can’t be overlooked, and has become too large to leave to data governance strategies focused elsewhere. BI Ops is the answer.
Want to explore BI Ops? Schedule a demo with Datalogz to learn more.
