Here’s to a great 2024! Given the stormy horizon for 2023, it’s a lot easier to be optimistic if we fast forward 12 months. With no such time machine available, we cast a sober eye on 2023 and all it brings: war, inflation… recession?

Despite these macro challenges, Eckerson Group’s team of experts sees great opportunities for innovation in how enterprises manage and analyze data. The following predictions build on the time-tested trends of tool convergence, data sharing, and improved governance on multiple fronts. 

1. Data Mesh and Data Fabric Converge 

As companies implement data mesh and data fabric approaches, their experiences reveal little reason to choose one over the other. The proponents of each approach sometimes give the impression that they’re mutually exclusive. However, they are more complementary than competitive. Therefore, we will continue to see organizations combine principles from both approaches. They will deliver data products (from data mesh) through an abstraction layer (from data fabric) that makes data easier for consumers to find and use. This will reduce the complexity of managing a self-serve data platform that relies on distributed and varied sources.

 - By Jay Piscioneri

Learn about our CDO TechVent, Data Mesh and Data Fabric Products: Tools to Deliver a Governed Self-Service Data Platform. Scheduled for August 2023. A Market Landscape report will follow in September.

2. Enterprises Apply FinOps to Cloud Analytics Projects

As enterprises shift analytics projects to the cloud, lured by the flexibility of renting IT resources on demand, they get surprisingly high monthly bills. Such surprises drive the rise of FinOps. This emerging discipline helps analysts, engineers, finance managers, and business owners collaborate to govern the cost of cloud-related projects. FinOps instills best practices, automates processes, and makes stakeholders accountable for the cost of cloud-related activities. In 2023, look for FinOps to become a required element of cloud-analytics projects. A rising number of DataOps tools, data observability tools, and cloud data platforms will support FinOps with cost analyzers and recommendation engines that help streamline cloud consumption.

Read our blog, The Rise of FinOps: Cost Governance for Cloud-Based Analytics.

- By Kevin Petrie

3. Data Governance Pivots to Focus on People

Leading enterprises acknowledge that we don’t really govern data; rather, we govern the behaviors of people that work with data. With that awareness, they recognize that the goals of governance — protection, quality, and value — are cultural issues. While policies and controls remain necessary, they are not at the forefront of governance. A shift from data-first to people-first governance reduces the focus on oversight and enforcement and prioritizes data ethics and data literacy. Intelligent oversight and enforcement needs are primarily addressed with smart technology, such as AI/ML capabilities in data catalogs, observability tools, etc. Data governance teams become the front line in shaping data culture. In the new year, we’ll see the early adopters of modern data governance, first steps in an evolution that will continue into 2024 and beyond.

- By Dave Wells

Read Dave Wells’ eBook, Building a Data Literacy Program: What, Why, and How. Also, check out our CDO TechVent, Data Governance Platforms, and accompanying Market Landscape report. These include a data governance maturity assessment and program methodology to address strategy, people, and process.

4. Data Catalogs Get Active and "Intelligent” 

To support self-service analytics, many companies have purchased data catalogs, but most have struggled to fully implement them. In response, data catalog vendors are transforming their products from static metadata repositories that people view to execution platforms that automate curation at scale and support direct data access and provisioning. The move to "active metadata" promises to leverage mountains of metadata to automate classification, trigger alerts, foster collaboration, transform data definitions into data quality rules, among other things. Vendors now call these next-generation products "data intelligence platforms." Expect most data catalog vendors to embrace this moniker in 2023. 

- By Wayne Eckerson

Learn about our CDO TechVent, Data Intelligence: Next-Generation Data Catalogs for Governing Enterprise Data. Scheduled for June 2023. A Market Landscape report will follow in July.

5. Data Marketplaces Find Their Place

All major cloud data platforms now offer (or soon will) a public data marketplace where buyers and sellers can exchange data, models, queries, and other data objects. Most data providers, however, will only offer their most generic data products on the public marketplaces. They will use them as promotional vehicles to lure potential buyers to their own proprietary data stores. These new specialized data stores will provide value-added services that minimize the friction of finding, evaluating, purchasing, integrating, and managing external data. 

In addition, most enterprises will tiptoe gradually into the data marketplace economy. Most are intrigued by the revenue implications of monetizing internal data assets. However, most need time to sort out legal and security implications of sharing data broadly. As a result, most will start their data sharing journey by implementing internal data marketplaces. This will give them experience creating and sharing data products among data domains and business units.  

- By Wayne Eckerson

Learn about our CDO TechVent, Data Sharing and Marketplaces: The New Frontier in Data. Scheduled for December 15, 2022. A Market Landscape report will follow in January 2023.

6. Data Products Crystallize as Marketplaces and Meshes Merge

There is great debate in the data mesh world about the definition of a data product. This is odd because emerging data marketplaces offered by Amazon, Snowflake, and others make it crystal clear what a data product is: it's any data object (usually a data set, query, or model)  packaged for distribution in a marketplace. The packaging consists of metadata descriptions, sample data, use cases, and possibly annotations and ratings by current and past users. In fact, a data marketplace is the missing element in the data mesh framework. Expect data mesh advocates to tout the value of their internal data marketplaces in 2023.

- By Wayne Eckerson

Learn about our CDO TechVent, Data Sharing and Marketplaces: The New Frontier in Data. Scheduled for December 15, 2022. A Market Landscape report will follow in January 2023.

7. Data Governance Gets Serious about Machine Learning

As enterprises build machine learning (ML) into more aspects of their business, they introduce new operational and compliance risks. ML models might automate bad decisions, breach privacy, or exhibit bias, which in turn disrupts operations, alienates customers, or invites regulatory scrutiny. Various governance-related tools are evolving to help control such risks. In 2023, look for a rising number of data governance platforms and catalogs to help organize, document, and apply policies to ML models alongside other data assets. Also look for data observability tools to help govern ML models by identifying data drift or quality issues. In addition, machine learning platforms will continue to merge with cloud data platforms, further blending the governance of ML models with traditional data sets and artifacts.

- By Kevin Petrie

Learn about our thought leadership report, Analytics and Data Governance in the Age of Machine Learning. Scheduled for July 2023.

8. Data Tools Improve User Experience with Metadata and Advanced Analytics

As enterprises push for increased adoption and proper usage of data catalogs, data prep tools, and BI tools, they need to enhance the user experience. Vendors will continue to make their tools more efficient, intuitive, and usable by leveraging metadata and advanced analytics. For example, Quaeris is an analytics tool that gives non-technical users a fast, easy way to solve everyday problems using natural language. Aurelius, meanwhile, provides an open-source user interface extension on top of the Apache Atlas Data Catalog so that the catalog can be used more intuitively by non-technical business users.

- By David Hendrawirawan

9. Data Pipeline Products Converge

After over a decade of balkanization of data pipeline functions, the pendulum is beginning to swing the other way toward convergence. Organizations no longer want to manage licensing and version updates of ETL, ELT, transformation, and reverse ETL products, or integrate discrete product functions into a complete pipeline solution. As a result, data platform providers such as Snowflake and AWS will continue to expand their data management offerings. They will address more customer operational pain points, simplify tool management, and capture more of enterprise data management budgets.

- By Jay Piscioneri

Learn about our CDO TechVent, Modern Data Pipelines: Accelerating, Automating, and Validating the Delivery of Complex Data. Scheduled for March 2023. A Market Landscape report will follow in April.

10. Data Management Tools Add Privacy Enhancing Technology (PET)

Data management vendors continue to strengthen privacy controls by adding privacy enhancing technology (PET) to a variety of products, including data observability tools, catalogs, and AI/ML platforms. Look for more such integration and development in 2023. PET features will include differential privacy, which obscures personally identifiable information (PII) with just the right amount of “noisy” data that meets specific privacy loss parameter; synthetic data, which mimics real private data; and homomorphic encryption, which performs computations on encrypted data without decrypting it. Partnerships also will contribute to the trend. For example, Synthesized now provides synthetic data to the BigID data intelligence platform so users can perform non-production activities without exposing PII.

- By David Hendrawirawan

Bonus prediction: AI Chatbots Spawn Startups… and Risks

OpenAI’s new prototype ChatGPT evokes wonder, fear, and perplexity from all that have tried it. This computational linguistics platform uses machine learning techniques to generate text in response to a stunning variety of questions. Its answers demonstrate good grammar, logic, and factual evidence—some of which are wrong. We asked who Wayne Eckerson is, and ChatGPT informed us without hesitation that Wayne wrote a book that someone else authored. No citations, no certainty level, just a request that we give the response a thumbs up or thumbs down. 

This experience exemplifies both the upside and downside of AI chatbots in 2023. The upside: they will save the time needed for professionals of all types to generate content. Many new startups will arise to address this opportunity. The downside: chatbots will create the need for more data quality and governance controls. The net productivity benefit will take time to develop. 

- By Kevin Petrie

George F. Will observed that “The future has a way of arriving announced. Its arrival is jolting when people have not prepared for it.” Here’s hoping that we’ve helped the Eckerson Group community prepare for an inventive and prosperous 2023.