🗄️Data Management: Data Quality (Part II: An Introduction)

Introduction

From time to time, I found myself in a meeting with people blaming the poor quality of data for the strange numbers appearing in a report, and often they were right. Sometimes you can see that without going too deep into details, other times you have to spend some considerable amount of time and effort and look over the raw data to identify the records that cause the unexpected variations. We know we have a problem with data quality, though why nobody is doing anything to improve it? Who’s responsible in the end for data quality? How data quality issues appear, how the data quality can be improved and by what means? When and where the data quality must be considered? Why do we have to consider data quality and in definitive what is data quality? Which are data quality’s characteristics or dimensions?

Without pretending to be an expert on data quality but using my and others’ experience and understanding in dealing with data quality, I will try to approach the above questions in several posts. And as usual, before answering more complex questions, inevitably we have to define first what data quality is.

What is Data Quality?

In several consulted sources, mainly books and blogs, data quality’s definition revolve around the 'fitness for use' syntagma (e.g. [1], [3], [4]). The definition makes data quality highly dependent on the context in which it’s considered because each request comes with different data quality expectations/requirements and data quality levels. Even more, the data could have acceptable quality in the legacy system(s), the system(s) used to store the data, though the data might not be appropriate for use in the current form. From this observation can be delimited two aspects of data quality – intrinsic data quality considering the creation and storage of data, and secondly contextual data quality referring to the degree the data meets users’ needs. 

Except these two aspects [2] considers two other aspects: the representation data quality referring to the "degree of care taken in the presentation and organization of information for users" and the accessibility data quality - "degree of freedom that users have to use data, define or refine the manner in which information is input, processed or presented to them" [2]. These four aspects provide a good framework for categorizing data quality and as can be seen, it targets the four coordinates of data – storage/creation, presentation, use and accessibility – however it ignores the fact data could be consumed also by machines, they needing to parse, understand and process the data. Therefore, could be added a fifth aspect – processing data quality, referring to the data readiness for consumption by machines, this aspect including the existence of metadata (e.g. data models, data mappings, etc.).


Why Data Quality?

Considering the growing dynamics and complexity of economic landscape, nowadays organization’s capacity of staying in business is highly dependent on decision making process’ accuracy, which at its turn is dependent on the reports’ quality used as support in decisions, and further on quality of data the reports are based on. If you have wrong data, you can’t expect the reports to show the correct numbers, same as you can’t expect a decision to be correct without having the actual situation correctly reflected even if occasionally the person taking the decision has a sound understanding of the business above the data level, or a good gut feeling. It is essential for an organization to have adequate data and reports that reflect the current state of art at any time and at the requested level of detail.

In this context is discussed about poor data quality, good data quality, high data quality, or even adequate/inadequate data quality. Paraphrasing data quality’s definition could be said that data are of high quality if they are 'fit for use' [1], and have poor data quality otherwise, though in life things are not black or white, but exist many other nuances in between, dependent on user’s perception and needs. For example, could be sufficient in one million records to have only one value that makes it impossible for a data set to be processed by a software package, and even if the data quality equates with a Six Sigma level, it’s not enough for this case. At the opposite side, a data set with a huge number of defects could be processed without problems and the outcome could be close to what it’s expected, now depends also on the nature of defects, which are the attributes affected and how they affect the result.

Poor data quality reduces considerably the value of data, impeding organizations to exploit the data at their full potential, a fact that limits their strategic competitive advantage. Poor data quality has a domino effect, often involving rework, delays in production, operations and decision making, a decrease in customers’ satisfaction reflected in revenues and lost opportunities, and so on. Unfortunately, the costs of poor data quality are difficult to quantify, they include the cost for finding and fixing the errors, the cost of wasted effort, lost production, and lost opportunity, plus the ones involved with wrong decision making. The same issue affects the costs involved in achieving high data quality levels, the easiest to quantify being the costs directly involved with the data quality initiatives. It's not in scope to detail these aspects, though is worth mentioning them to highlight why many organizations ignore the quality of their data.

How do Data Quality issues appear?

No system is bullet proof, no matter how good it was designed there will always be a chance for errors to appear, even if it’s the software itself, the people, the processes, standards, or procedures. A primary reason for data issues is caused by the weak validation implemented in the systems, either in the UI or in any other data entry points, by the errors that escaped the vigilant eyes of testers, by the (lack of) flexibility/robustness of applications, data models in general and data types in particular. On a secondary plane but not less important is the existence and adherence to processes, standards, or procedures, of whether they are known and respected by users, or whether they cover all the scenarios occurring in the daily business activity.

Who’s responsible for Data Quality?

As everybody in an organization is responsible for quality, the same applies also for data quality, from data entry to decision making everybody’s working with data directly or indirectly, and they are also theoretically responsible for it. From data life-cycle perspective there are the people who are entering the data in the various systems, the ones who are preparing the data for further usage, the ones using the data, and the ones managing the data quality initiatives (e.g. quality leaders, black belts, champions). The first category needs to take the ownership over the data, maintain them adequately, stick to the processes, standards and procedures defined, and together with the other users are responsible for identifying the issues in data, mitigate them and find solutions.

The data users are maybe the broadest category, ranging from simple employees using the data for daily activities to executives. Because often the data they are working with are aggregated at various levels, unless they have a good overview of the business or flexible reports that gives them access also to the row data it will be almost impossible for them to realize the actual data quality. Often the people preparing the data are the ones discovering the issues in data, though this happens by hazard, observing abnormalities in the data.

The people managing the data quality initiatives are hardly working with the data, though together with the other type of executives they are responsible for modeling an organization and creating a (data) quality culture.

How can be the Data Quality improved and by what means?

It takes time, resources, consolidated effort, adequate policies, and management’s involvement to reach an acceptable level of data quality. As already stressed data quality can be improved by creating robust processes, standards, or procedures, and sticking to them, creating a culture for quality in general and for data quality in particular, but rather they are the outcome of data quality initiatives.

The first step is recognizing that data quality in the organization as a problem and secondly identifying the nature of the issues eventually by conducting a data quality assessment, though this involve the definition of data quality (business) rules, rules against the quality of data will be checked, establishing an assessment, cleaning, and reporting framework for this purpose. Because improvement is the word of the day, could be conducted Six Sigma projects to identify issues’ root causes and the solutions to address them, monitor issues resolution and whether the found solutions fulfill the intended goals. The use of (Lean) Six Sigma methodology allows introducing quality concepts in the organization, making people aware of the premises and implications of data quality, and in addition it offers a receipt for success.

Depending on the volume of data and the number of issues found, data quality improvement might occur within several iterations, and when the volume of issues is more than organization can handle, then it makes sense to prioritize and fix the issues that impact the organization the most, following to address the remaining issues in time. There will also be situations in which is needed to redesign the existing processes, modify the legacy systems to limit the data entry issues or even start new projects in order fill the gaps.

When must be Data Quality considered?

Data quality comes in discussion especially when the data related issues impacts visibly the business, whether something is done to correct them or not that’s something dependent on whether an organization has in place a data quality vision, policy and adequate mechanisms for assessing and addressing the data quality on a regular basis, or whether the people or departments impacted by the issues take the problem in their hands and do something about it.

Another context in which data quality topic appears is during migration, conversion, and integration between systems, when the system(s) into which the data are supposed to be loaded, the destination system(s), come(s) with a set of rules/constraints requesting the data to be in a certain format/formatting, match specific data types and other data storage related constraints. Even if the data from the legacy system(s) are considered of high quality, in the mentioned activities the data quality is considered against the constraints imposed by the destination system.

What needs to be stressed is that data quality is not one time endeavor but a continuous and consolidated effort that could span along the life cycle of the systems storing the data. How often data quality assessments must be conducted? It depends on data growth and perceived quality, organizations’ data quality policy/vision and the degree they are respected, organization overall maturity and culture, availability of resources and the different usage of data.

Where is Data Quality considered?

Data quality is generally considered against the legacy systems the data following to be cleaned in them, though when data quality is judged against external requests, the data could be cleaned in Excel files or in any other environments designed for this kind of tasks, each scenario having its pluses and minuses, sometimes being preferred a hybrid solution between the two alternatives. Most of the time it makes sense to clean the data in the legacy system because in general it minimizes the overall effort, the data are “actual” and reflected in the other systems using the data. 

Excel or any other similar tools could prove to be better environments for cleaning the data for one time requests, when is needed to do fast mass updates or is not intended/possible to modify the legacy data, however the probability of making mistakes is quite high because the validation implemented in the legacy systems are not available anymore or must be enforced, multiple copies of the same data could exist in case the data needs to be clean by multiple users, while the synchronization between legacy data and cleaned data could be problematic.

There are situations in which the data are not available in any of the systems available in place and there is no adequate possibility to store the new data in them, thus being requested eventually to extend one or more systems (aka data enrichment), solution that could be quite expensive, an alternative being to store the data in ad-hoc solutions (e.g. Excel, MS Access), following to merge the existing data islands (aka data integration) before loading them into the destination system(s). When dealing with a relatively small number of records the data could be entered manually in the destination system itself, though there are all kind of constraints that need to be considered, therefore it is safer and more appropriate to load the data as contiguous sets.

Previous Post <<||>> Next Post

Written: Dec-2009, Last Reviewed: Mar-2024

References:
[1] Joseph M Juran & A Blanton Godfrey (1999) Juran’s Quality Handbook, 5th Ed.
[2] Coral Calero (2008). Handbook of Research on Web Information Systems Quality
[3] US Census Bureau (2006) Definition of Data Quality: Census Bureau Principle, version 1.3. [Online] Available from: http://www.census.gov/quality/P01-0_v1.3_Definition_of_Quality.pdf (Accessed: 24 December 2009).
[4] OCDQ (Obsessive-Compulsive Data Quality) Blog (2009) The Data-Information Continuum. [Online] Available from: http://www.ocdqblog.com/home/the-data-information-continuum.html (Accessed: 24 December 2009)

🔢Arkady Maydanchik - Collected Quotes

"Data cleansing is dangerous mainly because data quality problems are usually complex and interrelated. Fixing one problem may create many others in the same or other related data elements." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"Data quality program is a collection of initiatives with the common objective of maximizing data quality and minimizing negative impact of the bad data. [...] objective of any data quality program is to ensure that data quality docs not deteriorate during conversion and consolidation projects, Ideally, we would like to do even more and use the opportunity to improve data quality since data cleansing is much easier to perform before conversion than afterwards." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"Databases rarely begin their life empty. More often the starting point in their lifecycle is a data conversion from some previously exiting data source. And by a cruel twist of fate, it is usually a rather violent beginning. Data conversion usually takes the better half of new system implementation effort and almost never goes smoothly." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"[...] data conversion is the most difficult part of any system implementation. The error rate in a freshly populated new database is often an order of magnitude above that of the old system from which the data is converted. As a major source of the data problems, data conversion must be treated with the utmost respect it deserves." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"Equally critical is to include data quality definition and acceptable quality benchmarks into the conversion specifications. No product design skips quality specifications. including quality metrics and benchmarks. Yet rare data conversion follows suit. As a result, nobody knows how successful the conversion project was until data errors get exposed in the subsequent months and years. The solution is to perform comprehensive data quality assessment of the target data upon conversion and compare the results with pre-defined benchmarks." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"More and more data is exchanged between the systems through real-time (or near real-time) interfaces. As soon as the data enters one database, it triggers procedures necessary to send transactions to Other downstream databases. The advantage is immediate propagation of data to all relevant databases. Data is less likely to be out-of-sync. [...] The basic problem is that data is propagated too fast. There is little time to verify that the data is accurate. At best, the validity of individual attributes is usually checked. Even if a data problem can be identified. there is often nobody at the other end of the line to react. The transaction must be either accepted or rejectcd (whatever the consequences). If data is rejected, it may be lost forever!" (Arkady Maydanchik, "Data Quality Assessment", 2007)

"Much data in databases has a long history. It might have come from old 'legacy' systems or have been changed several times in the past. The usage of data fields and value codes changes over time. The same value in the same field will mean totally different thing in different records. Knowledge or these facts allows experts to use the data properly. Without this knowledge, the data may bc used literally and with sad consequences. The same is about data quality. Data users in the trenches usually know good data from bad and can still use it efficiently. They know where to look and what to check. Without these experts, incorrect data quality assumptions are often made and poor data quality becomes exposed." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"The big part of the challenge is that data quality does not improve by itself or as a result of general IT advancements. Over the years, the onus of data quality improvement was placed on modern database technologies and better information systems. [...] In reality, most IT processes affect data quality negatively, Thus, if we do nothing, data quality will continuously deteriorate to the point where the data will become a huge liability." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"The corporate data universe consists of numerous databases linked by countless real-time and batch data feeds. The data continuously move about and change. The databases are endlessly redesigned and upgraded, as are the programs responsible for data exchange. The typical result of this dynamic is that information systems get better, while data deteriorates. This is very unfortunate since it is the data quality that determines the intrinsic value of the data to the business and consumers. Information technology serves only as a magnifier for this intrinsic value. Thus, high quality data combined with effective technology is a great asset, but poor quality data combined with effective technology is an equally great liability." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"The greatest challenge in data conversion is that actual content and structure of the source data is rarely understood. More often data transformation algorithms rely on the theoretical data definitions and data models, Since this information is usually incomplete, outdated, and incorrect, the converted data look nothing like what is expected. Thus, data quality plummets. The solution is to precede conversion with extensive data profiling and analysis. In fact, data quality after conversion is in direct (or even exponential) relation with the amount of knowledge about actual data you possess. Lack of in-depth analysis will guarantee significant loss of data quality." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"The main tool of a data quality assessment professional is a data quality rule - a constraint that validates a data element or a relationship between several data elements and can be implemented in a computer program. [...] The solution relies on the design and implementation of hundreds and thousands of such data quality rules, and then using them to identify all data inconsistencies. Miraculously, a well-designed and fine-tuned collection of rules will identify a majority Of data errors in a fraction or time compared with manual validation. In fact, it never takes more than a few months to design and implement the rules and produce comprehensive error reports, What is even better, the same setup can be reused over and over again to reassess data quality periodically with minimal effort." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"Using data quality rules brings comprehensive data quality assessment from fantasy world to reality. However, it is by no means simple, and it takes a skillful skipper to navigate through the powerful currents and maelstroms along the way. Considering the volume and structural complexity of a typical database, designing a comprehensive set of data quality rules is a daunting task. The number of rules will often reach hundreds or even thousands. When some rules are missing, the results of the data quality assessment can be completely jeopardized, Thus the first challenge is to design all rules and make sure that they indeed identify all or most errors." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"While we might attempt to identify and correct most data errors, as well as try to prevent others from entering the database, the data quality will never be perfect. Perfection is practically unattainable in data quality as with the quality of most other products. In truth, it is also unnecessary since at some point improving data quality becomes more expensive than leaving it alone. The more efficient our data quality program, the higher level of quality we will achieve- but never will it reach 100%. However, accepting imperfection is not the same as ignoring it. Knowledge of the data limitations and imperfections can help use the data wisely and thus save time and money, The challenge, of course, is making this knowledge organized and easily accessible to the target users. The solution is a comprehensive integrated data quality meta data warehouse." (Arkady Maydanchik, "Data Quality Assessment", 2007)

🔢Robert Hawker - Collected Quotes

"[...] a conceptual data model [...] is system-agnostic and is a diagrammatic business representation of how different types of data are associated with one another in the organization." (Robert Hawker, "Practical Data Quality", 2023)

"A data quality rule is logic that is applied to each row of a dataset, which can determine whether the row of data is correct or incorrect. Correct data is deemed to have passed the rule, and incorrect data is deemed to have failed the rule – hence, the term failed data [...]" (Robert Hawker, "Practical Data Quality", 2023)

"Correction of data in the secondary source is not recommended. However, it is important to recognize that sometimes, secondary source fixes are required." (Robert Hawker, "Practical Data Quality", 2023)

"Data discovery is the process where an organization obtains an understanding of which data matters the most and identifies challenges with that data. The outcome of data discovery is that the scope of a data quality initiative should be clear and data quality rules can be defined." (Robert Hawker, "Practical Data Quality", 2023)

"Data profiling assesses a set of data and provides information on the values, the length of strings, the level of completeness, and the distribution patterns of each column." (Robert Hawker, "Practical Data Quality", 2023)

"Data quality rules are only effective if they are tightly scoped. Generic rules tend to produce a lot of unwanted failed records, and business users start to ignore the results. Once business users lose faith in what they see from a data quality tool, it is hard to restore engagement." (Robert Hawker, "Practical Data Quality", 2023)

"Every data quality initiative is different, and senior stakeholders at different organizations will have different needs." (Robert Hawker, "Practical Data Quality", 2023)

"If an organization had a single overall data quality key performance indicator (KPI), then it might be appropriate to put a greater weighting on those rules which would impact regulatory compliance. A lack of regulatory compliance is a risk to the very existence of organizations like these, and therefore, a greater weighting might be needed." (Robert Hawker, "Practical Data Quality", 2023)

"It rarely makes sense to aim for what people might consider perfect data (every record is complete, accurate, and up to date). The investment required is usually prohibitive, and the gains made for the last 1% of data quality improvement effort become far too marginal." (Robert Hawker, "Practical Data Quality", 2023)

"In truth, no one knows how much bad data quality costs a company – even companies with mature data quality initiatives in place, who are measuring hundreds of data points for their quality struggle to accurately measure quantitative impact. This is often a deal-breaker for senior leaders when trying to get approval for a budget for data quality work. Data quality initiatives often seek substantial budgets and are up against projects with more tangible benefits." (Robert Hawker, "Practical Data Quality", 2023)

"Momentum is important in data quality initiatives. If an issue is problematic, even where the priority is high, it can be better to move on to an issue that can be progressed efficiently." (Robert Hawker, "Practical Data Quality", 2023)

"Most data quality issues will re-occur if the root cause is not fully understood [...]" (Robert Hawker, "Practical Data Quality", 2023)

"Organizations will always only have a limited amount of resources available to remediate data. It will almost certainly not be possible to tackle all the issues at the same time. Therefore, prioritization is key to ensuring that the most value is generated from the available resources." (Robert Hawker, "Practical Data Quality", 2023)

"Successful organizations try to put a holistic data culture in place. Everyone is educated on the basics of looking after data and the importance of having good data. They consider what they have learned when performing their day-to-day tasks. This is often referred to as the promotion of good data literacy." (Robert Hawker, "Practical Data Quality", 2023)

"The biggest mistake that can be made in a data quality initiative is focusing on the wrong data. If you fix data that does not impact a critical business process or drive important decisions, your initiative simply will not make the difference that you want it to." (Robert Hawker, "Practical Data Quality", 2023)

"The data should be monitored in the source, it should be corrected in the source, and it should then feed the secondary source(s) with high-quality data that can be used without workarounds. The reduction in workarounds will make the data engineers, scientists, and data visualization specialists much more productive." (Robert Hawker, "Practical Data Quality", 2023)

"The level of data quality in an organization is the extent to which data can be used for its intended purposes."  (Robert Hawker, "Practical Data Quality", 2023)

"Start with a business strategy. Too many organizations start their data quality initiative by looking at the details of the data and trying to see 'what is wrong with it'. The right approach is to understand what the business is trying to achieve and to work out where data issues might impede this. It ensures that data quality work will be truly impactful." (Robert Hawker, "Practical Data Quality", 2023)

⛏️Data Management: Data Quality Management [DQM] (Definitions)

[Total Data Quality Management:] "An approach that manages data proactively as the outcome of a process, a valuable asset rather than the traditional view of data as an incidental by-product." (Karolyn Kerr, "Improving Data Quality in Health Care", 2009)

"The application of total quality management concepts and practices to improve data and information quality, including setting data quality policies and guidelines, data quality measurement (including data quality auditing and certification), data quality analysis, data cleansing and correction, data quality process improvement, and data quality education." (DAMA International, "The DAMA Dictionary of Data Management", 2011)

"Data Quality Management (DQM) is about employing processes, methods, and technologies to ensure the quality of the data meets specific business requirements." (Mark Allen & Dalton Cervo, "Strategy, Scope, and Approach" [in "Multi-Domain Master Data Management"], 2015)

"DQM is the management of company data in a manner aware of quality. It is a sub-function of data management and analyzes, improves and assures the quality of data in the company. DQM includes all activities, procedures and systems to achieve the data quality required by the business strategy. Among other things, DQM transfers approaches for the management of quality for physical goods to immaterial goods like data." (Boris Otto & Hubert Österle, "Corporate Data Quality", 2015)

"Data quality management (DQM) is a set of practices aimed at improving and maintaining the quality of data across a company’s business units." (altexsoft) [source]

"Data quality management is a set of practices that aim at maintaining a high quality of information. DQM goes all the way from the acquisition of data and the implementation of advanced data processes, to an effective distribution of data. It also requires a managerial oversight of the information you have." (Data Pine) [source]

"Data quality management is a setup process, which is aimed at achieving and maintaining high data quality. Its main stages involve the definition of data quality thresholds and rules, data quality assessment, data quality issues resolution, data monitoring and control." (ScienceSoft) [source]

"Data quality management is the act of ensuring suitable data quality." (Xplenty) [source]

"Data quality management provides a context-specific process for improving the fitness of data that’s used for analysis and decision making. The goal is to create insights into the health of that data using various processes and technologies on increasingly bigger and more complex data sets." (SAS) [source]

"Data quality management (DQM) refers to a business principle that requires a combination of the right people, processes and technologies all with the common goal of improving the measures of data quality that matter most to an enterprise organization." (BMC) [source]

"Put most simply, data quality management is the process of reviewing and updating your customer data to minimize inaccuracies and eliminate redundancies, such as duplicate customer records and duplicate mailings to the same address." (EDQ) [source]

🎯William Smith - Collected Quotes

"Achieving a gold standard for data quality at ingestion involves a multifaceted approach: defining explicit schemas and contracts, implementing rigorous input validation reflecting domain semantics, supporting immediate rejection or secure quarantine of low-quality data, and embedding these capabilities into high-throughput, low-latency pipelines. This first line of defense not only prevents downstream data pollution but also establishes an enterprise-wide culture and infrastructure aimed at preserving data trust from the point of entry onward." (William Smith, "Great Expectations for Modern Data Quality: The Complete Guide for Developers and Engineers", 2025)

"Accuracy denotes the degree to which data correctly represents the real-world entities or events to which it refers." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"At its core, data quality encompasses multiple dimensions-including accuracy, completeness, consistency, timeliness, validity, uniqueness, and relevance-that require rigorous assessment and control. The progression from traditional data management practices to cloud-native, real-time, and federated ecosystems introduces both challenges challenges and opportunities for embedding quality assurance seamlessly across the entire data value chain." (William Smith, "Great Expectations for Modern Data Quality: The Complete Guide for Developers and Engineers", 2025) 

"At its core, observability rests on three fundamental pillars: metrics, logs, and traces. In the context of data systems, these pillars translate into quantitative measurements (such as data volume, processing latency, and schema changes), detailed event records (including data pipeline execution logs and error messages), and lineage traces that map the flow of data through interconnected processes. Together, they enable a granular and multidimensional understanding of data system behavior, facilitating not just detection but also rapid root-cause analysis." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Completeness refers to the extent to which required data attributes or records are present in a dataset." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Consistency signifies the absence of conflicting data within or across sources. As data ecosystems become distributed and federated, ensuring consistency transcends simple referential integrity checks."(William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Data drift refers to shifts in the statistical properties or distributions of incoming data compared to those observed during training or baseline establishment. Common variants include covariate drift (changes in feature distributions), prior probability drift (changes in class or label proportions), and concept drift (changes in the relationship between features and targets)." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Data governance establishes the overarching policies, standards, and strategic directives that define how data assets are to be managed across the enterprise. This top-level framework sets the boundaries of authority, compliance requirements, and key performance indicators for data quality." (William Smith, "Great Expectations for Modern Data Quality: The Complete Guide for Developers and Engineers", 2025) 

"Data Lakes embrace a schema-on-read approach, storing vast volumes of raw or lightly processed data in native formats with minimal upfront constraints. This design significantly enhances ingestion velocity and accommodates diverse, unstructured, or semi-structured datasets. However, enforcing data quality at scale becomes more complex, as traditional static constraints are absent." (William Smith, "Great Expectations for Modern Data Quality: The Complete Guide for Developers and Engineers", 2025) 

"Data mesh fundamentally reframes data governance and validation by distributing accountability to domain-oriented teams who act as custodians and producers of their respective data products. These teams possess intimate domain knowledge, which is essential for nuanced validation criteria that adapt to the semantics, context, and evolution of their datasets. By treating datasets as first-class products with clear ownership, interfaces, and service-level objectives, data mesh encourages autonomous validation workflows embedded directly within the domains where data originates and is consumed." (William Smith, "Great Expectations for Modern Data Quality: The Complete Guide for Developers and Engineers", 2025)

"Data quality insights generated through automated profiling and baseline analysis are only as valuable as their visibility and actionability within the broader organizational decision-making context." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Data quality verification, when executed as a set of static, invariant rules, often fails to accommodate the inherent fluidity of real-world datasets and evolving analytical contexts. To ensure robustness and relevance, quality checks must evolve beyond static constraints, incorporating adaptability driven by metadata, runtime information, and domain-specific business logic. This transformation enables the development of dynamic and context-aware validation systems capable of offering intelligent, self-tuning quality enforcement with reduced false positives and operational noise." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Effective management of data quality at scale requires a clear delineation of organizational roles and operational frameworks that ensure accountability, consistency, and continuous improvement. Central to this structure are the interrelated concepts of data governance, data stewardship, and operational ownership. Each serves distinct, yet complementary purposes in embedding responsibility within technology platforms, business processes, and organizational culture." (William Smith, "Great Expectations for Modern Data Quality: The Complete Guide for Developers and Engineers", 2025) 

"Establishing a comprehensive observability architecture necessitates a systematic approach that spans the entirety of the data pipeline, from initial telemetry collection to actionable insights accessible by diverse stakeholders. The core objective is to unify distributed data sources - metrics, logs, traces, and quality signals - into a coherent framework that enables rapid diagnosis, continuous monitoring, and strategic decision-making." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Governance sets the strategic framework, stewardship bridges strategy with execution, and operational ownership grounds responsibility within systems and processes. Advanced organizations achieve sustainable data quality by establishing clear roles, defined escalation channels, embedded tooling, standardized processes, and a culture that prioritizes data excellence as a collective, enforceable mandate." (William Smith, "Great Expectations for Modern Data Quality: The Complete Guide for Developers and Engineers", 2025)  

"Modern complex organizations increasingly confront the challenge of ensuring data quality at scale without centralizing validation activities into a single bottlenecked team. The data mesh paradigm and federated controls emerge as pivotal architectural styles and organizational patterns that enable decentralized, self-serve data quality validation while preserving coherence and reliability across diverse data products." (William Smith, "Great Expectations for Modern Data Quality: The Complete Guide for Developers and Engineers", 2025)

"Observability [...] requires that systems be instrumented to expose rich telemetry, enabling ad hoc exploration and hypothesis testing regarding system health. Thus, observability demands design considerations at the architecture level, insisting on standardization of instrumentation, consistent metadata management, and tight integration across data processing, storage, and orchestration layers." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Quality gates embody a comprehensive strategy for continuous data assurance by enforcing hierarchical checks, asserting dynamic SLAs, and automating compliance decisions grounded in explicit policies. Their architecture and operationalization directly address the complex interplay between technical robustness and regulatory compliance, ensuring that only trusted data permeates downstream systems." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Robust access control forms the cornerstone of observability system security. At the core lies the principle of least privilege, wherein users and service identities are granted the minimal set of permissions required to perform their designated tasks. This principle substantially reduces the attack surface by minimizing unnecessary access and potential lateral movement paths within the system. Implementing least privilege necessitates fine-grained role-based access control (RBAC) models tailored to organizational roles and operational workflows. RBAC configurations should be explicit regarding the scopes and data domains accessible to each role, avoiding overly broad privileges." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Relevance gauges the appropriateness of data for the given analytical or business context. Irrelevant data, though possibly accurate and complete, can introduce noise and degrade model performance or decision quality." (William Smith, "Great Expectations for Modern Data Quality: The Complete Guide for Developers and Engineers", 2025) 

"Robust methodologies to measure and prioritize data quality dimensions involve composite metrics and scoring systems that combine quantitative indicators-such as error rates, completeness percentages, latency distributions-with qualitative assessments from domain experts." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"The architecture of a robust data quality framework hinges fundamentally on three interconnected pillars: open standards, extensible application programming interfaces (APIs), and interoperable protocols. These pillars collectively enable the seamless exchange, validation, and enhancement of data across diverse platforms and organizational boundaries." (William Smith, "Great Expectations for Modern Data Quality: The Complete Guide for Developers and Engineers", 2025) 

"The data swamp anti-pattern arises from indiscriminate ingestion of uncurated data, which rapidly dilutes data warehouse utility and complicates quality monitoring." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"The selection of KPIs should be driven by a rigorous alignment with business objectives and user requirements. This mandates close collaboration with stakeholders spanning data scientists, operations teams, compliance officers, and executive sponsors." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Timeliness captures the degree to which data is available when needed and reflects the relevant time frame of the underlying phenomena." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Uniqueness ensures that each entity or event is captured once and only once, preventing duplication that can distort analysis and decision-making." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025

"Validity reflects whether data conforms to the syntactic and semantic rules predefined for its domain." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

🗃️Data Management: Data Quality (Just the Quotes)

"[...] it is a function of statistical method to emphasize that precise conclusions cannot be drawn from inadequate data." (Egon S Pearson & H Q Hartley, "Biometrika Tables for Statisticians" Vol. 1, 1914)

"Not even the most subtle and skilled analysis can overcome completely the unreliability of basic data." (Roy D G Allen, "Statistics for Economists", 1951)

"The enthusiastic use of statistics to prove one side of a case is not open to criticism providing the work is honestly and accurately done, and providing the conclusions are not broader than indicated by the data. This type of work must not be confused with the unfair and dishonest use of both accurate and inaccurate data, which too commonly occurs in business. Dishonest statistical work usually takes the form of: (1) deliberate misinterpretation of data; (2) intentional making of overestimates or underestimates; and (3) biasing results by using partial data, making biased surveys, or using wrong statistical methods." (John R Riggleman & Ira N Frisbee, "Business Statistics", 1951)

"Data are of high quality if they are fit for their intended use in operations, decision-making, and planning." (Joseph M Juran, 1964)

"There is no substitute for honest, thorough, scientific effort to get correct data (no matter how much it clashes with preconceived ideas). There is no substitute for actually reaching a correct chain of reasoning. Poor data and good reasoning give poor results. Good data and poor reasoning give poor results. Poor data and poor reasoning give rotten results." (Edmund C Berkeley, "Computers and Automation", 1969)

"Detailed study of the quality of data sources is an essential part of applied work. [...] Data analysts need to understand more about the measurement processes through which their data come. To know the name by which a column of figures is headed is far from being enough." (John W Tukey, "An Overview of Techniques of Data Analysis, Emphasizing Its Exploratory Aspects", 1982)

"We have found that some of the hardest errors to detect by traditional methods are unsuspected gaps in the data collection (we usually discovered them serendipitously in the course of graphical checking)." (Peter Huber, "Huge data sets", Compstat '94: Proceedings, 1994)

"Data obtained without any external disturbance or corruption are called clean; noisy data mean that a small random ingredient is added to the clean data." (Nikola K Kasabov, "Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering", 1996)

"Probability theory is a serious instrument for forecasting, but the devil, as they say, is in the details - in the quality of information that forms the basis of probability estimates." (Peter L Bernstein, "Against the Gods: The Remarkable Story of Risk", 1996)

"Unfortunately, just collecting the data in one place and making it easily available isn’t enough. When operational data from transactions is loaded into the data warehouse, it often contains missing or inaccurate data. How good or bad the data is a function of the amount of input checking done in the application that generates the transaction. Unfortunately, many deployed applications are less than stellar when it comes to validating the inputs. To overcome this problem, the operational data must go through a 'cleansing' process, which takes care of missing or out-of-range values. If this cleansing step is not done before the data is loaded into the data warehouse, it will have to be performed repeatedly whenever that data is used in a data mining operation." (Joseph P Bigus,"Data Mining with Neural Networks: Solving business problems from application development to decision support", 1996)

"If the data is usually bad, and you find that you have to gather some data, what can you do to do a better job? First, recognize what I have repeatedly said to you, the human animal was not designed to be reliable; it cannot count accurately, it can do little or nothing repetitive with great accuracy. [...] Second, you cannot gather a really large amount of data accurately. It is a known fact which is constantly ignored. It is always a matter of limited resources and limited time. [...] Third, much social data is obtained via questionnaires. But it a well documented fact the way the questions are phrased, the way they are ordered in sequence, the people who ask them or come along and wait for them to be filled out, all have serious effects on the answers."  (Richard Hamming, "The Art of Doing Science and Engineering: Learning to Learn", 1997)

"Blissful data consist of information that is accurate, meaningful, useful, and easily accessible to many people in an organization. These data are used by the organization’s employees to analyze information and support their decision-making processes to strategic action. It is easy to see that organizations that have reached their goal of maximum productivity with blissful data can triumph over their competition. Thus, blissful data provide a competitive advantage." (Margaret Y Chu, "Blissful Data", 2004)

"Let’s define dirty data as: ‘… data that are incomplete, invalid, or inaccurate’. In other words, dirty data are simply data that are wrong. […] Incomplete or inaccurate data can result in bad decisions being made. Thus, dirty data are the opposite of blissful data. Problems caused by dirty data are significant; be wary of their pitfalls."  (Margaret Y Chu, "Blissful Data", 2004)

"Processes must be implemented to prevent bad data from entering the system as well as propagating to other systems. That is, dirty data must be intercepted at its source. The operational systems are often the source of informational data; thus dirty data must be fixed at the operational data level. Implementing the right processes to cleanse data is, however, not easy." (Margaret Y Chu, "Blissful Data", 2004)

"Equally critical is to include data quality definition and acceptable quality benchmarks into the conversion specifications. No product design skips quality specifications. including quality metrics and benchmarks. Yet rare data conversion follows suit. As a result, nobody knows how successful the conversion project was until data errors get exposed in the subsequent months and years. The solution is to perform comprehensive data quality assessment of the target data upon conversion and compare the results with pre-defined benchmarks." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"Much data in databases has a long history. It might have come from old 'legacy' systems or have been changed several times in the past. The usage of data fields and value codes changes over time. The same value in the same field will mean totally different thing in different records. Knowledge or these facts allows experts to use the data properly. Without this knowledge, the data may bc used literally and with sad consequences. The same is about data quality. Data users in the trenches usually know good data from bad and can still use it efficiently. They know where to look and what to check. Without these experts, incorrect data quality assumptions are often made and poor data quality becomes exposed." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"The big part of the challenge is that data quality does not improve by itself or as a result of general IT advancements. Over the years, the onus of data quality improvement was placed on modern database technologies and better information systems. [...] In reality, most IT processes affect data quality negatively, Thus, if we do nothing, data quality will continuously deteriorate to the point where the data will become a huge liability." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"While we might attempt to identify and correct most data errors, as well as try to prevent others from entering the database, the data quality will never be perfect. Perfection is practically unattainable in data quality as with the quality of most other products. In truth, it is also unnecessary since at some point improving data quality becomes more expensive than leaving it alone. The more efficient our data quality program, the higher level of quality we will achieve- but never will it reach 100%. However, accepting imperfection is not the same as ignoring it. Knowledge of the data limitations and imperfections can help use the data wisely and thus save time and money, The challenge, of course, is making this knowledge organized and easily accessible to the target users. The solution is a comprehensive integrated data quality meta data warehouse." (Arkady Maydanchik, "Data Quality Assessment", 2007)

"Achieving a high level of data quality is hard and is affected significantly by organizational and ownership issues. In the short term, bandaging problems rather than addressing the root causes is often the path of least resistance." (Cindi Howson, "Successful Business Intelligence: Secrets to making BI a killer App", 2008)

"Communicate loudly and widely where there are data quality problems and the associated risks with deploying BI tools on top of bad data. Also advise the different stakeholders on what can be done to address data quality problems - systematically and organizationally. Complaining without providing recommendations fixes nothing." (Cindi Howson, "Successful Business Intelligence: Secrets to making BI a killer App", 2008)

"Data quality is such an important issue, and yet one that is not well understood or that excites business users. It’s often perceived as being a problem for IT to handle when it’s not: it’s for the business to own and correct." (Cindi Howson, "Successful Business Intelligence: Secrets to making BI a killer App", 2008)

"Depending on the extent of the data quality issues, be careful about where you deploy BI. Without a reasonable degree of confidence in the data quality, BI should be kept in the hands of knowledge workers and not extended to frontline workers and certainly not to customers and suppliers. Deploy BI in this limited fashion as data quality issues are gradually exposed, understood, and ultimately, addressed. Don’t wait for every last data quality issue to be resolved; if you do, you will never deliver any BI capabilities, business users will never see the problem, and quality will never improve." (Cindi Howson, "Successful Business Intelligence: Secrets to making BI a killer App", 2008)

"Our culture, obsessed with numbers, has given us the idea that what we can measure is more important than what we can't measure. Think about that for a minute. It means that we make quantity more important than quality." (Donella Meadows, "Thinking in Systems: A Primer", 2008)

"The data architecture is the most important technical aspect of your business intelligence initiative. Fail to build an information architecture that is flexible, with consistent, timely, quality data, and your BI initiative will fail. Business users will not trust the information, no matter how powerful and pretty the BI tools. However, sometimes it takes displaying that messy data to get business users to understand the importance of data quality and to take ownership of a problem that extends beyond business intelligence, to the source systems and to the organizational structures that govern a company’s data." (Cindi Howson, "Successful Business Intelligence: Secrets to making BI a killer App", 2008)

"Many new data scientists tend to rush past it to get their data into a minimally acceptable state, only to discover that the data has major quality issues after they apply their (potentially computationally intensive) algorithm and get a nonsense answer as output. (Sandy Ryza, "Advanced Analytics with Spark: Patterns for Learning from Data at Scale", 2009)

"Access to more information isn’t enough - the information needs to be correct, timely, and presented in a manner that enables the reader to learn from it. The current network is full of inaccurate, misleading, and biased information that often crowds out the valid information. People have not learned that 'popular' or 'available' information is not necessarily valid." (Gene Spafford, 2010)

"Are data quality and data governance the same thing? They share the same goal, essentially striving for the same outcome of optimizing data and information results for business purposes. Data governance plays a very important role in achieving high data quality. It deals primarily with orchestrating the efforts of people, processes, objectives, technologies, and lines of business in order to optimize outcomes around enterprise data assets. This includes, among other things, the broader cross-functional oversight of standards, architecture, business processes, business integration, and risk and compliance. Data governance is an organizational structure that oversees the compliance and standards of enterprise data." (Neera Bhansali, "Data Governance: Creating Value from Information Assets", 2014)

"Data governance is about putting people in charge of fixing and preventing data issues and using technology to help aid the process. Any time data is synchronized, merged, and exchanged, there have to be ground rules guiding this. Data governance serves as the method to organize the people, processes, and technologies for data-driven programs like data quality; they are a necessary part of any data quality effort." (Neera Bhansali, "Data Governance: Creating Value from Information Assets", 2014)

"Having data quality as a focus is a business philosophy that aligns strategy, business culture, company information, and technology in order to manage data to the benefit of the enterprise. Data quality is an elusive subject that can defy measurement and yet be critical enough to derail a single IT project, strategic initiative, or even an entire company." (Neera Bhansali, "Data Governance: Creating Value from Information Assets", 2014)

"Accuracy and coherence are related concepts pertaining to data quality. Accuracy refers to the comprehensiveness or extent of missing data, performance of error edits, and other quality assurance strategies. Coherence is the degree to which data - item value and meaning are consistent over time and are comparable to similar variables from other routinely used data sources." (Aileen Rothbard, "Quality Issues in the Use of Administrative Data Records", 2015)

"How good the data quality is can be looked at both subjectively and objectively. The subjective component is based on the experience and needs of the stakeholders and can differ by who is being asked to judge it. For example, the data managers may see the data quality as excellent, but consumers may disagree. One way to assess it is to construct a survey for stakeholders and ask them about their perception of the data via a questionnaire. The other component of data quality is objective. Measuring the percentage of missing data elements, the degree of consistency between records, how quickly data can be retrieved on request, and the percentage of incorrect matches on identifiers (same identifier, different social security number, gender, date of birth) are some examples." (Aileen Rothbard, "Quality Issues in the Use of Administrative Data Records", 2015)

"When we find data quality issues due to valid data during data exploration, we should note these issues in a data quality plan for potential handling later in the project. The most common issues in this regard are missing values and outliers, which are both examples of noise in the data." (John D Kelleher et al, "Fundamentals of Machine Learning for Predictive Data Analytics: Algorithms, worked examples, and case studies", 2015)

"A popular misconception holds that the era of Big Data means the end of a need for sampling. In fact, the proliferation of data of varying quality and relevance reinforces the need for sampling as a tool to work efficiently with a variety of data, and minimize bias. Even in a Big Data project, predictive models are typically developed and piloted with samples." (Peter C Bruce & Andrew G Bruce, "Statistics for Data Scientists: 50 Essential Concepts", 2016)

"Metadata is the key to effective data governance. Metadata in this context is the data that defines the structure and attributes of data. This could mean data types, data privacy attributes, scale, and precision. In general, quality of data is directly proportional to the amount and depth of metadata provided. Without metadata, consumers will have to depend on other sources and mechanisms." (Saurabh Gupta et al, "Practical Enterprise Data Lake Insights", 2018)

"The quality of data that flows within a data pipeline is as important as the functionality of the pipeline. If the data that flows within the pipeline is not a valid representation of the source data set(s), the pipeline doesn’t serve any real purpose. It’s very important to incorporate data quality checks within different phases of the pipeline. These checks should verify the correctness of data at every phase of the pipeline. There should be clear isolation between checks at different parts of the pipeline. The checks include checks like row count, structure, and data type validation." (Saurabh Gupta et al, "Practical Enterprise Data Lake Insights", 2018)

"Are your insights based on data that is accurate and reliable? Trustworthy data is correct or valid, free from significant defects and gaps. The trustworthiness of your data begins with the proper collection, processing, and maintenance of the data at its source. However, the reliability of your numbers can also be influenced by how they are handled during the analysis process. Clean data can inadvertently lose its integrity and true meaning depending on how it is analyzed and interpreted." (Brent Dykes, "Effective Data Storytelling: How to Drive Change with Data, Narrative and Visuals", 2019)

"First, from an ethos perspective, the success of your data story will be shaped by your own credibility and the trustworthiness of your data. Second, because your data story is based on facts and figures, the logos appeal will be integral to your message. Third, as you weave the data into a convincing narrative, the pathos or emotional appeal makes your message more engaging. Fourth, having a visualized insight at the core of your message adds the telos appeal, as it sharpens the focus and purpose of your communication. Fifth, when you share a relevant data story with the right audience at the right time (kairos), your message can be a powerful catalyst for change." (Brent Dykes, "Effective Data Storytelling: How to Drive Change with Data, Narrative and Visuals", 2019)

"The one unique characteristic that separates a data story from other types of stories is its fundamental basis in data. [...] The building blocks of every data story are quantitative or qualitative data, which are frequently the results of an analysis or insightful observation. Because each data story is formed from a collection of facts, each one represents a work of nonfiction. While some creativity may be used in how the story is structured and delivered, a true data story won’t stray too far from its factual underpinnings. In addition, the quality and trustworthiness of the data will determine how credible and powerful the data story is." (Brent Dykes, "Effective Data Storytelling: How to Drive Change with Data, Narrative and Visuals", 2019)

"Data is dirty. Let's just get that out there. How is it dirty? In all sorts of ways. Misspelled text values, date format problems, mismatching units, missing values, null values, incompatible geospatial coordinate formats, the list goes on and on." (Ben Jones, "Avoiding Data Pitfalls: How to Steer Clear of Common Blunders When Working with Data and Presenting Analysis and Visualizations", 2020) 

"Bad data makes bad models. Bad models instruct people to make ineffective or harmful interventions. Those bad interventions produce more bad data, which is fed into more bad models." (Cory Doctorow, "Machine Learning’s Crumbling Foundations", 2021)

"[...] data mesh introduces a fundamental shift that the owners of the data products must communicate and guarantee an acceptable level of quality and trustworthiness - specific to their domain - as an intrinsic characteristic of their data product. This means cleansing and running automated data integrity tests at the point of the creation of a data product." (Zhamak Dehghani, "Data Mesh: Delivering Data-Driven Value at Scale", 2021)

"Ensure you build into your data literacy strategy learning on data quality. If the individuals who are using and working with data do not understand the purpose and need for data quality, we are not sitting in a strong position for great and powerful insight. What good will the insight be, if the data has no quality within the model?" (Jordan Morrow, "Be Data Literate: The data literacy skills everyone needs to succeed", 2021)

"[...] the governance function is accountable to define what constitutes data quality and how each data product communicates that in a standard way. It’s no longer accountable for the quality of each data product. The platform team is accountable to build capabilities to validate the quality of the data and communicate its quality metrics, and each domain (data product owner) is accountable to adhere to the quality standards and provide quality data products." (Zhamak Dehghani, "Data Mesh: Delivering Data-Driven Value at Scale", 2021)

"Bad data is costly to fix, and it’s more costly the more widespread it is. Everyone who has accessed, used, copied, or processed the data may be affected and may require mitigating action on their part. The complexity is further increased by the fact that not every consumer will “fix” it in the same way. This can lead to divergent results that are divergent with others and can be a nightmare to detect, track down, and rectify." (Adam Bellemare, "Building an Event-Driven Data Mesh: Patterns for Designing and Building Event-Driven Architectures", 2023)

"Data has historically been treated as a second-class citizen, as a form of exhaust or by-product emitted by business applications. This application-first thinking remains the major source of problems in today’s computing environments, leading to ad hoc data pipelines, cobbled together data access mechanisms, and inconsistent sources of similar-yet-different truths. Data mesh addresses these shortcomings head-on, by fundamentally altering the relationships we have with our data. Instead of a secondary by-product, data, and the access to it, is promoted to a first-class citizen on par with any other business service." (Adam Bellemare, "Building an Event-Driven Data Mesh: Patterns for Designing and Building Event-Driven Architectures", 2023)

"In truth, no one knows how much bad data quality costs a company – even companies with mature data quality initiatives in place, who are measuring hundreds of data points for their quality struggle to accurately measure quantitative impact. This is often a deal-breaker for senior leaders when trying to get approval for a budget for data quality work. Data quality initiatives often seek substantial budgets and are up against projects with more tangible benefits." (Robert Hawker, "Practical Data Quality", 2023)

"The biggest mistake that can be made in a data quality initiative is focusing on the wrong data. If you fix data that does not impact a critical business process or drive important decisions, your initiative simply will not make the difference that you want it to." (Robert Hawker, "Practical Data Quality", 2023)

"The data should be monitored in the source, it should be corrected in the source, and it should then feed the secondary source(s) with high-quality data that can be used without workarounds. The reduction in workarounds will make the data engineers, scientists, and data visualization specialists much more productive." (Robert Hawker, "Practical Data Quality", 2023)

"The problem of bad data has existed for a very long time. Data copies diverge as their original source changes. Copies get stale. Errors detected in one data set are not fixed in duplicate ones. Domain knowledge related to interpreting and understanding data remains incomplete, as does support from the owners of the original data." (Adam Bellemare, "Building an Event-Driven Data Mesh: Patterns for Designing and Building Event-Driven Architectures", 2023)

"Data Lakes embrace a schema-on-read approach, storing vast volumes of raw or lightly processed data in native formats with minimal upfront constraints. This design significantly enhances ingestion velocity and accommodates diverse, unstructured, or semi-structured datasets. However, enforcing data quality at scale becomes more complex, as traditional static constraints are absent." (William Smith, "Great Expectations for Modern Data Quality: The Complete Guide for Developers and Engineers", 2025) 

"Data quality insights generated through automated profiling and baseline analysis are only as valuable as their visibility and actionability within the broader organizational decision-making context." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"Data quality verification, when executed as a set of static, invariant rules, often fails to accommodate the inherent fluidity of real-world datasets and evolving analytical contexts. To ensure robustness and relevance, quality checks must evolve beyond static constraints, incorporating adaptability driven by metadata, runtime information, and domain-specific business logic. This transformation enables the development of dynamic and context-aware validation systems capable of offering intelligent, self-tuning quality enforcement with reduced false positives and operational noise." (William Smith, "Soda Core for Modern Data Quality and Observability: The Complete Guide for Developers and Engineers", 2025)

"LLM-centric workloads change everything. Now the raw material is heterogeneous text, code, images, audio, and chat logs whose value depends on semantic richness - that is, the informational value of the content - rather than a rigid structure. Pipelines must tokenize, chunk, embed, and version this content; store it in vector indexes for similarity search; and apply filters for personally identifiable information, toxicity, and licensing constraints. Instead of ETL jobs, teams run continuous ingestion and re-embedding loops so that RAG systems stay fresh, and they log every prompt–response pair so that the inputs and outputs can be evaluated and improve the future performance of this system. Data quality in this context is judged by grounding, factuality, and bias metrics - attributes that require automated red-teaming and human-in-the-loop (HITL) review rather than the data structure violation checks of the past." (Abi Aryan, "LLMOps: Managing Large Language Models in Production", 2025)

"The AI adoption challenges that businesses are facing share a common thread: AI is not a magic solution. Many businesses fall into the trap of expecting AI to solve their problems quickly, without fully understanding its limitations or properly integrating it into their operations. Data quality plays a crucial role in the success of AI systems; if the data are biased, incomplete, or outdated, the results will reflect those shortcomings. In addition, the human element is vital. AI should augment, not replace, human intelligence in areas that require empathy, creativity, or complex decision-making, such as dealing with the many experts right across the organisation. Additionally, AI systems need to be adaptable, accounting for external factors such as cultural differences, market changes, and unpredictable human behaviours." (Alan Watkins & G C Cooke, "Smarter than You Winning in Business with Superintelligent AI", 2026)

"Errors using inadequate data are much less than those using no data at all." (Charles Babbage)

⛏️Data Management: Data Quality (Definitions)

"Data are of high quality if they are fit for their intended use in operations, decision-making, and planning." (Joseph M Juran, 1964)

"[…] data has quality if it satisfies the requirements of its intended use. It lacks quality to the extent that it does not satisfy the requirement. In other words, data quality depends as much on the intended use as it does on the data itself. To satisfy the intended use, the data must be accurate, timely, relevant, complete, understood, and trusted." (Jack E Olson, "Data Quality: The Accuracy Dimension", 2003)

"A set of measurable characteristics of data that define how well data represents the real-world construct to which it refers." (Alex Berson & Lawrence Dubov, "Master Data Management and Customer Data Integration for a Global Enterprise", 2007)

"The state of completeness, validity, consistency, timeliness and accuracy that makes data appropriate for a specific use." (Keith Gordon, "Principles of Data Management", 2007)

"Deals with data validation and cleansing services (to ensure relevance, validity, accuracy, and consistency of the master data), reconciliation services (aimed at helping cleanse the master data of duplicates as part of consistency), and cross-reference services (to help with matching master data across multiple systems)." (Martin Oberhofer et al,"Enterprise Master Data Management", 2008)

"A set of data properties (features, parameters, etc.) describing their ability to satisfy user’s expectations or requirements concerning data using for information acquiring in a given area of interest, learning, decision making, etc." (Juliusz L Kulikowski, "Data Quality Assessment", 2009)

"Assessment of the cleanliness, accuracy, and reliability of data." (Laura Reeves, "A Manager's Guide to Data Warehousing", 2009)

"A set of measurable characteristics of data that define how well the data represents the real-world construct to which it refers." (Alex Berson & Lawrence Dubov, "Master Data Management and Data Governance", 2010)

"This term refers to whether an organization’s data is reliable, consistent, up to date, free of duplication, and can be used efficiently across the organization." (Tony Fisher, "The Data Asset", 2009)

"A set of measurable characteristics of data that define how well the data represents the real-world construct to which it refers." (Alex Berson & Lawrence Dubov, "Master Data Management and Data Governance", 2010)

"The degree of data accuracy, accessibility, relevance, time-liness, and completeness." (Linda Volonino & Efraim Turban, "Information Technology for Management" 8th Ed., 2011)

"The degree of fitness for use of data in particular application. Also the degree to which data conforms to data specifications as measured in data quality dimensions. Sometimes used interchangeably with information quality." (John R Talburt, "Entity Resolution and Information Quality", 2011) 

"The degree to which data is accurate, complete, timely, consistent with all requirements and business rules, and relevant for a given use." (DAMA International, "The DAMA Dictionary of Data Management", 2011)

"Contextual data quality considers the extent to which data are applicable (pertinent) to the task of the data user, not to the context of representation itself. Contextually appropriate data must be relevant to the consumer, in terms of timeliness and completeness. Dimensions include: value-added, relevancy, timeliness, completeness, and appropriate amount of data (from the Wang & Strong framework.)" (Laura Sebastian-Coleman, "Measuring Data Quality for Ongoing Improvement ", 2012)

"Intrinsic data quality denotes that data have quality in their own right; it is understood largely as the extent to which data values are in conformance with the actual or true values. Intrinsically good data is accurate, correct, and objective, and comes from a reputable source. Dimensions include: accuracy objectivity, believability, and reputation (from the Wang & Strong framework)." (Laura Sebastian-Coleman, "Measuring Data Quality for Ongoing Improvement ", 2012)

"Representational data quality indicates that the system must present data in such a way that it is easy to understand (represented concisely and consistently) so that the consumer is able to interpret the data; understood as the extent to which data is presented in an intelligible and clear manner. Dimensions include: interpretability, ease of understanding, representational consistency, and concise representation (rom the Wang & Strong framework)." (Laura Sebastian-Coleman, "Measuring Data Quality for Ongoing Improvement ", 2012)

"The level of quality of data represents the degree to which data meets the expectations of data consumers, based on their intended use of the data." (Laura Sebastian-Coleman, "Measuring Data Quality for Ongoing Improvement", 2013) 

"The relative value of data, which is based on the accuracy of the knowledge that can be generated using that data. High-quality data is consistent, accurate, and unambiguous, and it can be processed efficiently." (Jim Davis & Aiman Zeid, "Business Transformation: A Roadmap for Maximizing Organizational Insights", 2014)

"The properties of data embodied by the “Five C’s”: clean, consistent, conformed, current, and comprehensive." (Daniel Linstedt & W H Inmon, "Data Architecture: A Primer for the Data Scientist", 2014)

"The degree to which data in an IT system is complete, up-to-date, consistent, and (syntactically and semantically) correct." (Tilo Linz et al, "Software Testing Foundations, 4th Ed", 2014)

"A measure for the suitability of data for certain requirements in the business processes, where it is used. Data quality is a multi-dimensional, context-dependent concept that cannot be described and measured by a single characteristic, but rather various data quality dimensions. The desired level of data quality is thereby oriented on the requirements in the business processes and functions, which use this data [...]" (Boris Otto & Hubert Österle, "Corporate Data Quality", 2015)

"[...] characteristics of data such as consistency, accuracy, reliability, completeness, timeliness, reasonableness, and validity. Data-quality software ensures that data elements are represented in a consistent way across different data stores or systems, making the data more trustworthy across the enterprise." (Judith S Hurwitz, "Cognitive Computing and Big Data Analytics", 2015)

"Refers to the accuracy, completeness, timeliness, integrity, and acceptance of data as determined by its users." (Gregory Lampshire, "The Data and Analytics Playbook", 2016)

"A measure of the useableness of data. An ideal dataset is accurate, complete, timely in publication, consistent in its naming of items and its handling of e.g. missing data, and directly machine-readable (see data cleaning), conforms to standards of nomenclature in the field, and is published with sufficient metadata that users can easily understand, for example, who it is published by and the meaning of the variables in the dataset." (Open Data Handbook) 

"Refers to the level of 'quality' in data. If a particular data store is seen as holding highly relevant data for a project, that data is seen as quality to the users." (Solutions Review)

"the processes and techniques involved in ensuring the reliability and application efficiency of data. Data is of high quality if it reliably reflects underlying processes and fits the intended uses in operations, decision making and planning." (KDnuggets)

"The narrow definition of data quality is that it's about data that is missing or incorrect. A broader definition is that data quality is achieved when a business uses data that is comprehensive, consistent, relevant and timely." (Information Management)

"Data Quality refers to the accuracy of datasets, and the ability to analyse and create actionable insights for other users." (experian) [source]

"Data quality refers to the current condition of data and whether it is suitable for a specific business purpose." (Xplenty) [source]

🔢Rupa Mahanti - Collected Quotes

"A data model is a formal organized representation of real-world entities, focused on the definition of an object and its associated attributes and relationships between the entities. Data models should be designed consistently and coherently. They should not only meet requirements, but should also enable data consumers to better understand the data." (Rupa Mahanti, "Data Quality: Dimensions, Measurement, Strategy, Management, and Governance", 2019)

"Bad data are expensive: my best estimate is that it costs a typical company 20% of revenue. Worse, they dilute trust - who would trust an exciting new insight if it is based on poor data! And worse still, sometimes bad data are simply dangerous; look at the damage brought on by the financial crisis, which had its roots in bad data." (Rupa Mahanti, "Data Quality: Dimensions, Measurement, Strategy, Management, and Governance", 2019)

"Conformity, or validity, means the data comply with a set of internal or external standards or guidelines or standard data definitions, including metadata definitions. Comparison between the data items and metadata enables measuring the degree of conformity." (Rupa Mahanti, "Data Quality: Dimensions, Measurement, Strategy, Management, and Governance", 2019)

"Data-intensive projects generally involve at least one person who understands all the nuances of the application, process, and source and target data. These are the people who also know about all the abnormalities in the data and the workarounds to deal with them, and are the experts. This is especially true in the case of legacy systems that store and use data in a manner it should not be used. The knowledge is not documented anywhere and is usually inside the minds of the people. When the experts leave, with no one having a true understanding of the data, the data are not used properly and everything goes haywire." (Rupa Mahanti, "Data Quality: Dimensions, Measurement, Strategy, Management, and Governance", 2019)

"Data are collections of facts, such as numbers, words, measurements, observations, or descriptions of real-world objects, phenomena, or events and their attributes. Data are qualitative when they contain descriptive information, and quantitative when they contain numerical information." (Rupa Mahanti, "Data Quality: Dimensions, Measurement, Strategy, Management, and Governance", 2019)

"Data migration generally involves the transfer of data from an existing data source to a new database or to a new schema within the same database. [...] Data migration projects deal with the migration of data from one data structure to another data structure, or data transformed from one platform to another platform with modified data structure." (Rupa Mahanti, "Data Quality: Dimensions, Measurement, Strategy, Management, and Governance", 2019)

"Data quality is the capability of data to satisfy the stated business, system, and technical requirements of an enterprise. Data quality is an insight into or an evaluation of data’s fitness to serve their purpose in a given context. Data quality is accomplished when a business uses data that are complete, relevant, and timely. The general definition of data quality is 'fitness for use', or more specifically, to what extent some data successfully serve the purposes of the user." (Rupa Mahanti, "Data Quality: Dimensions, Measurement, Strategy, Management, and Governance", 2019)

"Lack of a standard process to address business requirements and business process improvements, poorly designed and implemented business processes that result in lack of training, coaching, and communication in the use of the process, and unclear definition of subprocess or process ownership, roles, and responsibilities have an adverse impact on data quality." (Rupa Mahanti, "Data Quality: Dimensions, Measurement, Strategy, Management, and Governance", 2019)

"The degree of data quality excellence that should be attained and sustained is driven by the criticality of the data, the business need and the cost and time to achieve the defined degree of data quality." (Rupa Mahanti, "Data Quality: Dimensions, Measurement, Strategy, Management, and Governance", 2019)

"To understand why data quality is important, we need to understand the categorization of data, the current quality of data and how is it different from the quality of manufacturing processes, the business impact of bad data and cost of poor data quality, and possible causes of data quality issues." (Rupa Mahanti, "Data Quality: Dimensions, Measurement, Strategy, Management, and Governance", 2019)

🗄️Data Management: Data Cleansing (Part I: An Introduction)

Data Management Series

Even if often Data Cleansing or Data Scrubbing is considered synonymous to Data Cleaning, it actually includes the Data Validation and Data Cleaning process, where Data Validation is the process of checking data quality against a set of rules (referred as data quality rules) in a given set of data, and Data Cleaning or Data Correction is the process of correcting the data quality issues identified in Data Validation. Data Cleansing consists in general of the following steps:

Step 1: Defining the scope

The scope definition resumes at the identification of elements, attributes and the records that need to be cleaned. If is intended only to improve overall data quality and nothing else, the focus will be mainly on the data elements and attributes perceived as having low data quality. For data conversion the elements in scope are defined based on the requirements of the destination system and overall business requirements are defined the attributes in scope, an exercise that needs to be performed is the data mapping between the destination on one side and the legacy system(s) and the other data sources (e.g. Excel, MS Access files) on the other side.

From case to case in scope for cleansing might be all the records available in the legacy system(s) for a given set of data elements and attributes, or only a fraction of them, usually the focus is mainly the active and open records, and eventually some historical records. By active records are referred those records not cancelled or disabled, either master and transactional data , by open records the (transactional) records not marked as closed, records on which an action is still pending (e.g. pending Invoice Receipt, Payment, etc.), while by historical records are referred the already closed transactional data (e.g. Purchase Orders, Sales Orders, etc.).

Given the importance of master data for the business they are often the starting point for cleaning, they being also the data elements with the longest issue resolution, the transactional data being in theory easier to clean. The two types of data need to be synchronized the referent data related to the transactional data in scope needs to be loaded too.

Step 2: Defining the data quality rules and error types.

The data quality rules are defined for the attributes in scope and the data quality dimensions considered that apply, typically data accuracy, completeness, conformity, consistency, duplication and referential integrity. The issues are eventually further categorized and prioritized.

Step 3: Validating the data quality rules

The data quality rules are validated against the actual data sets, the output of the rules are consolidated in reports that follow to be interpreted. The number of errors are reported and stored for historical trending in case several iterations are needed for data cleaning.

Step 4: Identifying the data quality issues.

The reports are reviewed by the data owners, the issues being identified and the rules validated, further steps being performed in order to identify the root causes for the discovered issues. This usually leads to the calibration of data quality rules as new scenarios (e.g. exceptions, new error types) can be discovered.

Step 5: Correcting the data quality issues.

The issues are mitigated and corrected, issues’ resolution time, the time needed to correct the issues, is directly dependent on the volume of issues, their complexity, the number of resources allocated to the task, on whether the issues are corrected in the source or on a copy of the data, on whether the cleansing is done manually or specific software tools are used. Specific Data Cleaning software tools can help decreasing considerably the volume of manual work involved during the process, being preferred to use automatic and semiautomatic data cleansing methods whenever is possible.

Step 6: Validating the changes against the same data quality rules.

Once the issues are corrected the data sets are validated against the same rules used at Step 3, the Steps 3 to 5 being repeated until the expected data quality level is achieved. As already highlighted in a previous post, Data Cleansing is not a one-time endeavor, but an iterative process, new iterations being requested by changes occurred in rules or in scope, by the sequencing/scheduling of cleansing work or by reporting frequency (centered or not along the important milestones).

Step 7: Documenting/Reporting error results and validation outcomes, if needed.

Starting with data elements, attributes, and the logic used to select the records in scope for cleaning, and ending with the data quality reports with the issues and the improvement trends it makes sense to document everything and summarize it in a set of best practices and lesson learned sessions.

Step 8: Modify processes/procedures to reduce issues occurrence.

There are chances to discover that the errors found during validation could be avoided by modifying the existing processes, procedures or standards, or enforcing new validation rules in the legacy systems. When the changes are complex might be started even individual projects to address them.