Data Management and Analytics: An Introduction for CIOs and New Data Leaders

AI Advisor · Free Tool

Technology Landscape Advisor

Describe your technology challenge and get an AI-generated landscape analysis: relevant technology categories, key vendors (commercial and open source), recommended architecture patterns, and a curated shortlist — all tailored to your industry, organization size, and constraints.

Vendor-neutral analysis

Architecture patterns

Downloadable Word report

Analyze My Landscape View All AI Advisors

Data management is the work of getting data captured, stored, secured, defined and kept accurate. Analytics is the work of using that data to answer questions and make decisions. The two are one supply chain: analytics consumes what data management produces, and every analytics result is only as good as the data behind it. DAMA International, the professional body for the field, defines data management as "the development, execution, and supervision of plans, policies, programs, and practices that deliver, control, protect, and enhance the value of data and information assets throughout their lifecycles." This guide covers what each side includes, how they connect, who does the work, and the order in which to build it.

What data management covers: the DMBOK knowledge areas

DAMA publishes the Data Management Body of Knowledge (DAMA-DMBOK). DAMA says the DMBOK "is organized around 11 core knowledge areas, with data governance at the center." The names below are the ones on DAMA's DMBOK infographics page. The one-line descriptions are ours.

Knowledge area	What it covers in practice
Data Governance and Stewardship	Who decides what the data means, who may use it, and who is accountable when it is wrong
Data Architecture	The blueprint: which systems hold which data, and how data moves between them
Data Modeling and Design	The structure of the data: entities, attributes, keys and relationships
Data Storage and Operations	Running the databases: backup, recovery, performance, retention
Data Security	Access control, encryption, privacy and breach prevention
Data Integration and Interoperability	Moving and combining data between systems: ETL, ELT, APIs, streaming
Document and Content Management	Unstructured content such as contracts, email and images
Reference and Master Data	Shared code lists, and one agreed record for customers, products, suppliers and locations
Data Warehousing and Business Intelligence	Collecting data for analysis and presenting it in reports and dashboards
Metadata Management	Data about data: definitions, owners, lineage, where each field came from
Data Quality Management	Measuring and fixing accuracy, completeness, consistency and timeliness

The same infographics page also covers data handling ethics, big data and data science, and data management maturity assessment. Governance sits in the middle of DAMA's wheel diagram because every other area needs decisions about ownership and rules.

The current text is the DMBOK second edition with a 2024 revision. DAMA calls that revision a maintenance release and says "no material changes were made to the overarching data management framework, knowledge areas, or their definitions." It added currency as a data quality dimension and AI governance and ethics practices. A DMBOK 3.0 project is under way to incorporate "AI, cloud, and modern data platforms," with publication targeted for 2027.

You do not need all eleven areas staffed on day one. Start with governance, integration, warehousing and quality, and add the rest as the need appears.

What analytics covers: four questions

Analytics is usually split into four types, each answering a harder question than the last.

Type	Question it answers	Example	What it needs from data management
Descriptive	What happened?	Revenue by region last quarter	Integrated, reconciled history and agreed metric definitions
Diagnostic	Why did it happen?	Revenue fell in one region because two large accounts churned	Detail-level data that can be drilled into and joined across systems
Predictive	What is likely to happen?	Which accounts are likely to churn next quarter	Long, consistent history and features that mean the same thing over time
Prescriptive	What should we do?	Which retention offer to send each at-risk account	All of the above, plus data on actions taken and their outcomes

Each step depends on the one before it. A churn model trained on customer records that are duplicated across three systems will learn from the duplicates. Get descriptive reporting trusted before funding prediction.

How the two connect

Data travels from the systems that create it to the people and models that use it. The main stages are below.

Sources. ERP, CRM, billing, web and mobile apps, sensors, files from partners. These systems are built to run transactions, not to answer questions across the business.

Pipelines. Integration jobs copy data out of the sources. In ETL, data is transformed before it is loaded into the analytical store. In ELT, raw data is loaded first and transformed inside the store. Streaming moves events continuously instead of in batches. The trade-offs are covered in Building Data Pipelines That Scale.

The analytical store. A data warehouse holds structured, modeled data for reporting. A data lake holds raw files of any type at low cost. A lakehouse tries to combine the two. The 2021 CIDR paper by Databricks and academic co-authors that introduced the term defines a lakehouse as "a data management system based on low-cost and directly-accessible storage that also provides traditional analytical DBMS management and performance features such as ACID transactions, data versioning, auditing, indexing, caching, and query optimization." The paper's authors work for a lakehouse vendor. Read it as a design argument, not a neutral survey.

The semantic layer. This is where "revenue," "active customer" and "churn" get one definition each, written down once and used by every dashboard and query. Without it, two reports pull the same tables, apply different filters, and show the finance committee two numbers.

Consumption. BI dashboards, ad hoc queries, spreadsheets, data science notebooks, machine learning models and AI assistants that query data on a user's behalf.

Three disciplines run across every stage rather than sitting at one:

Governance sets who owns each data domain and which rules apply.
Quality checks data at the pipeline and in the store, and stops bad loads before they reach a dashboard.
Metadata records definitions, owners and lineage, so a user can see where a number came from and an engineer can see what a change will break.

How these stages are split among central and domain teams is a separate decision, covered in Data Mesh vs. Data Fabric. Which tools fill each stage is covered in The Modern Data Stack.

The people involved

Job titles vary by company. The responsibilities below appear in some form in most data organizations.

Role	Responsibility
Chief data officer or head of data	Owns the data strategy, budget and governance function
Data owner	A business person accountable for a data domain, such as customer or product
Data steward	Maintains definitions and quality rules for a domain day to day
Data architect	Designs the models, stores and flows
Data engineer or data integration developer	Builds and runs pipelines
Analytics engineer	Builds the modeled tables and metric definitions analysts query
BI analyst	Builds reports and dashboards and answers business questions
Data scientist	Builds statistical and machine learning models

DAMA's 2024 revision changed three of these. It now defines the data owner as "a business person accountable for decisions within a domain." It recognizes data scientists as "key data consumers, especially in analytics and integration," and replaced "Interface Developer" with "Data Integration Developer."

Fill the data owner role first. When IT owns the data, nobody with business authority settles disputes about what a field means, and the disputes return in every meeting.

The 80% figure, and what was actually measured

We could not trace the claim that data scientists spend 80% of their time cleaning data to a study that measured it (the nearest survey we found measured 45%).

The nearest primary source we could open is Anaconda's 2020 State of Data Science report. On page 11, respondents reported spending 19% of their time on data loading and 26% on data cleansing, 45% combined. Data visualization took 21%. Model selection, model training and scoring, and model deployment took 11%, 12% and 11%. That question had 1,099 answers. The survey ran from February 12 to April 20, 2020, and respondents were recruited through social media and Anaconda's email list. It describes people who chose to answer a data science vendor's survey, not data teams in general.

In that survey, preparation took a large share of analytical time, but not 80%. Quote 45% with its source, or quote no number.

DAMA's own What is Data Management page says "research shows that poor data quality can cost organizations up to 30% of their annual revenue." The page cites no research. We did not use the figure.

Build the foundation in the order the questions need it

Pick three to five decisions. Name the decisions the business wants better data for, such as pricing, inventory or customer retention. Every later step is scoped to them.
Assign owners for the data those decisions use. One business owner per domain, with a steward. Write down the definitions of the ten or twenty terms that matter most. Data Governance Framework sets out a 90-day starting sequence.
Integrate the sources into one analytical store. Warehouse or lakehouse, chosen for the workloads you have now. Load only what the chosen decisions need.
Put quality checks on the critical fields. Test completeness, uniqueness and freshness at load time. Alert the steward, not just the engineer.
Publish definitions and lineage. A catalog or semantic layer that lets an analyst find the agreed table and the agreed metric.
Deliver descriptive and diagnostic reporting. Retire the conflicting spreadsheets as each trusted report goes live.
Then add predictive and prescriptive work. Start with one model where the decision, the data and the measure of success are already clear.

Steps 2 to 5 produce no dashboard. If they are cut, the dashboards in step 6 show numbers nobody agrees on.

Frequently asked questions

What is the difference between data management and data analytics? Data management makes data available, accurate, secure and defined. Analytics uses that data to describe, explain, predict or recommend. Analytics depends on data management, not the other way around.

How many knowledge areas are in the DAMA DMBOK? Eleven, with data governance at the center, according to DAMA. The 2024 revision of the second edition did not change them. DMBOK 3.0 is targeted for 2027 and may.

Do we need a data warehouse before we can do analytics? You need one integrated, governed place to query. That can be a warehouse or a lakehouse. Reporting straight off operational systems works for one system and breaks when a question spans several.

Sources: DAMA International, "What is Data Management?", accessed October 22, 2026. DAMA International, "DAMA-DMBOK infographics", accessed October 22, 2026. DAMA International, "DAMA DMBOK 2.0 Revision: Maintenance Release", undated, describing the 2024 revision. DAMA International, "What Is DAMA International? A Guide for Data Professionals", April 14, 2026. DAMA International, "DAMA-DMBOK 3.0 Project", accessed October 22, 2026. Armbrust, Ghodsi, Xin and Zaharia, "Lakehouse: A New Generation of Open Platforms that Unify Data Warehousing and Advanced Analytics", CIDR 2021, section 3. Anaconda, 2020 State of Data Science, pages 3 and 11. The four analytics types are described here without attribution because we could not open a primary source for their origin. What would make this wrong, and when: DMBOK 3.0 is targeted for 2027. When it is published, re-check the knowledge-area table and the role definitions against it, and replace the edition paragraph.