Data Platform Value Propositions
For internal audiences, data platforms have four primary value propositions: metrics, decisions, trust, and governance.
If you provide a data platform for your clients or suppliers, your mileage may differ in a more formal product context. Because most data platforms serve internal audiences, those principles are held lightly when you do not need to solve for client segments, competitive differentiation, channels, or direct revenue.
However, value propositions remain front and center.
Metrics
Business stakeholders use internal data platforms to measure business activities using metrics such as key performance indicators (KPI).
A data platform provides value to decision-makers by encoding business rules into metrics that matter.
Data analysts work with teams in marketing, sales, finance, operations and product to define the metrics, to discover the data sources, and to validate alignment between the data creators and the decision-makers in how the data will be used.
Data engineers build data pipelines to automate the collection, standardization, cleansing, and quality controls for data that will be used for decisions. Because the decision context differs for each team, metrics for one team may filter or aggregate data differently than other teams.
Dashboards and reporting use a data platform to provide context-aware metrics specific to each team. Marketing seeks general trends, while sales & finance need to see the reality of daily sales, and operations wants to know the impacts on staffing. Different decision contexts require metrics that are unique to each.
Product management, design, and engineering teams use a data platform to study application usage, sometimes as aggregated patterns, and sometimes as detailed logs that compare before-and-after or capture alternatives as features are added to internal & client-facing applications.
Decisions
Many decisions use data in repeatable ways, and what matters most is consistency in metric definitions, how trends are presented, and how fresh the data content is. Visualization will match the stakeholders business context and data consumption preferences.
A data platform provides value to decision-makers through clarity, context, freshness, and customizability.
Detailed provenance is essential. A data platform needs to provide access to the underlying data when KPIs and dashboards show unexpected results.
Data platforms are also used by data scientists to carry out research, to test new hypotheses, and to experiment with new questions and new ways to answer them. However, please note that data platforms for internal audiences primarily contain data within the business as it exists today. This is a valuable to data scientists. However, they will also explore data outside the business, such as from industry competitors, equity partners, universities & governments. These sources are outside the scope of data platforms that serve internal business stakeholders.
Trust
To build and to keep the trust of decision-makers, a data platform needs to have the right data, sustained organizational support, and well-designed processes that guarantee accuracy and completeness.
A data platform provides value when decision-makers trust its insights and how it produces them.
Data owners of reference data that is used across departments need to know how their data is used, and they need curation and quality processes. When a data platform is introduced, it requires greater transparency and more effective communications when curated reference data changes.
Changes to data pipelines, to reports and dashboards, and to the underlying technology stack need the same controls as any internal software application. Release schedules, advanced notice of changes, validation after changes, and schema changes require version control and the ability to revert to a prior version. In some businesses, the data content itself may need versioning and the ability to revert to prior data.
Building new data pipelines for a data platform is an opportunity to introduce data content checking so that both engineers and decision-makers can receive automated alerts when data values are missing or null, or when a range or distribution of values fails the asserted expectations.
A data platform is easier to trust when the data underlying reports can be viewed by report users, and when the chain of transformations in the data pipeline are also visible. Although reports & dashboards may filter or aggregate, the data platform also shows the unfiltered and uncleansed source data to troubleshoot unexpected results.
Governance
Data platforms reduce risk by strengthening data access and authorization in order to implement compliance procedures. Standardizing how data access is granted via role-based access control (RBAC) provides the foundation for audit-friendly process documentation.
A data platform provides value to decision-makers when it complies with government regulations and satisfies external expectations.
Adding a semantic layer that describes why data is kept and what it represents permits the tagging of personal identifiable information (PII), protected healthcare information (PHI), and credit information coming from the payment card industry (PCI). Within the US, regulations of these data categories are the primary reasons to restrict access.
Adding federated data structures such as queryable data stores with partitions held in more than one cloud-hosted data center provides a way to keep data local to the country which regulates its sovereignty. Documentation within a data platform can provide organizational guidance on how consent is captured and who handles individual data inquiry and removal requests.
Although US law is slowly catching up to international law, firms that do business outside the US already recognize the need for informed consent, data sovereignty, secured access, and the inquiry & removal of data about individual data subjects. The data platform is the path to implement these risk-reduction process controls in order to demonstrate compliance with government regulations and to meet expectations of clients & data subjects.
