Kathy McDermott has more than 20 years of experience in the financial services industry. She joined Cutter Associates in 2011 and has authored a number of Cutter Research reports on a wide range of topics. She spent ten years as a consultant for a variety of asset management firms, with a focus on business analysis for front and middle office projects. Kathy was previously the senior business analyst, equity trading systems, at Wellington Management, specializing in the firm’s proprietary electronic and basket trading applications. Earlier in her career, Kathy worked for Thomson Financial Services (now Thomson Reuters), supporting FirstCall and PORTIA clients in Hong Kong, Japan, Singapore, and Thailand, and later went on to manage PORTIA implementations. She also worked at LongView (now Linedata) as an account manager and then product manager of electronic trading. Kathy earned her bachelor of arts in mathematics from Hamilton College.
Recent research assignments and publications include the following:
Alternative Data and the Expanding Universe of Investment Information
Behavioral Analysis in Investment Management
Client Reporting Systems and Practices
Data Access for Business Users
Managed Data Services
Market Data Administration
Mastering Client, Account, and Product Data
Order Management Systems
Outsourced Trading: Has the Time Come?
Reference Data Management Solutions
Aug 28, 2023
With insights from data-driven analytics improving their decision-making, investment management firms are exploring the various ways that they can harness the power of data to achieve their business objectives. Supporting self-service analytics by placing data in the hands of business users is becoming a goal for data-driven organizations.
One of the key areas to supporting self-service analytics is providing business users with access to critical business data. In Cutter’s 2023 Data Management Benchmarking Survey, we found that over half of firms (55%) do not provide business users with easy access to critical data and that business users frequently have to search for necessary data, run multiple reports and extracts to obtain the data, and then manually combine and analyze the data.
It’s a time-consuming and error-prone process to retrieve results, and a situation that grows increasingly difficult as firms add more data and types of data, and the number of teams and business users interested in data analytics rises.
Architecture Changes Support Data Analytics
At Cutter, we often talk about how a good data management practice hinges on a firm’s ability to execute on Cutter’s Four Cornerstones of Good Data Management (see below), and how executing each quadrant influences the success of a firm effectively executing on the other quadrants.
This article focuses on how improvements in the Architecture quadrant support the Analytics & Delivery quadrant. By example, we illustrate in today’s article how a firm’s ability to provide an architecture with data storage and tools that support data analytics underpins its ability (or inability) to support data access for business users.
Most of the statistics we reference come from our recent Data Management Benchmarking Survey in which firms answered a detailed questionnaire and were evaluated against Cutter’s Capability Model on whether their capabilities were optimized, managed, defined, or reactive. Cutter’s full2023 Data Management Benchmarking report is available here.
The issues that firms face in supporting their business users’ data access needs directly result from inefficiencies in their data architecture and the tools they make available to users. Establishing the following tools and data stores can substantially improve data access for business users and, as a result, self-service analytics:
A semantic layer to simplify access to core data
A data catalog or data dictionary designed for business users
An analytics data store
Implement a Semantic Layer
As a best practice, our Best Practices in Self-Service Data Analytics infographic specifies, “Deploy a semantic layer to consolidate and organize data from internal and external sources.” And for good reason ─ a semantic layer serves many purposes, including supporting self-service analytics and providing an abstraction layer that limits disruption when modernizing your data architecture. Before firms can analyze data and users can create their own analytics, they need access to data in a format they can use.
Unfortunately, the results of our Data Management Benchmarking Survey show that while investment management firms may understand the benefits of a semantic layer, most have yet to implement one. These results emphasize that firms still require manual effort from data consumers to combine data, which can act as a major stumbling block to achieving self-service analytics for business users.
Implement a Data Catalog or Data Dictionary
As both the volume and complexity of data increase, and more firms encourage self-service data analytics, it’s increasingly important for firms to provide business-oriented data catalogs or data dictionaries. Data cataloging products offer capabilities for the automation of metadata processes, such as metadata discovery and classification. These products help firms create and maintain a full inventory of data assets. They also help data consumers find the data they need, identify the source of the data, and understand the proper use.
While IT and data specialists traditionally used data dictionaries and catalogs, the advent of self-service analytics behooves firms to provide these tools to business users. Here, however, our benchmark survey results illustrate that most firms (44%) still only provide data catalogs to IT ─ or worse, they don’t have any data catalogs at all (28%).
Create an Analytics Data Store
Many firms have multiple data warehouses due to legacy reasons or different types of data. They might have an on-premises investment data warehouse and a cloud-based data warehouse that holds ESG data. Firms with multiple data repositories run the risk of replicating data in multiple places and creating issues in keeping data in sync in multiple locations. For self-service analytics, it’s inefficient for data consumers to look for data in multiple data stores, and it increases the risk of business users obtaining data from unauthorized sources. A semantic layer certainly helps in this situation, but so too can creating a consolidated analytics data store.
In comparison to implementing semantic layers or data catalogs, our benchmark survey participants show slightly greater maturity in creating analytics data stores, with just under 50% of participating firms in the upper half of Cutter’s Capability Model (leaving the majority firms in the lower half).
For firms that do not already have an analytics data store, building one in the cloud offers many benefits. Cloud data stores support the complexities of analytics processing because they include dynamic compute resources that adjust to the workload. These products typically offer prebuilt connections to analytics tools, programming languages, and visualization tools. Cloud providers such as AWS and Azure are known for both their data warehouse and analytics offerings. Snowflake has been a disruptor in this space for many reasons, especially because of its ease of implementation and data-share capabilities that allow firms to quickly onboard new datasets and keep data current without replicating it.
As firms increasingly transition data to cloud data stores, it’s even more important for them to implement a semantic layer that offers a layer of abstraction that masks data moving from one data store to another, without impacting business users’ access to that data. Firms that have experience with the volume of data that cloud data warehouses and data lakes can store know that implementing a data catalog becomes more important as the volume of data expands.
Firms today recognize that they are at a key opportunity point in their data analytics journey. In this article, we’ve concentrated on the impact that improvements in the Architecture quadrant can have on the Analytics & Delivery quadrant, specifically improving self-service data analytics for the business user.
Of course, all of the quadrants of Cutter’s Four Cornerstones of Good Data Management need to work together. Depending on where firms find themselves in their data journey, they may experience similar improvements in data analytics by implementing centralized DataOps (for more information on DataOps, see our recent article:DataOps: The Key to Managing a Modern Data Platform) and further expanding data governance.