Why asset managers should care about their data architecture
1. Why asset managers should care about their data architecture
Data serves as the input for all your business information. Once niche topics, ESG and alternative data, data mining, and business intelligence have become important and omnipresent. However, businesses can only reap the value from data and data-driven activities if the underlying data architecture is fit for purpose. It is the data architecture, which ensures the current and future information demand can be met in a scalable and qualitatively satisfying manner.
Understanding the data architecture helps management to focus on the different areas where change is needed to fulfill their information needs. For ESG data as a prominent example, the difficulty lies less in its technical provisioning, but rather in its quality, relevance, comparability, and interpretation. For real-time pricing and market exposure data, the complexity resides in the data infrastructure itself, while for business intelligence the data's availability in a single source of truth is more likely to be key.
A weak data architecture frequently requires IT changes at the entry point, or upstream systems of the organization's data infrastructure, before a new business or end user requirement can be met. Conversely, a good data architecture will ensure that you can focus on creating business value without lengthy IT projects, launch new products swiftly, and respond to client requests in a timely fashion.
2. Good data architecture is a differentiator, building everything yourself is not
It may sound like an IT topic, which is not connected to selling any products or creating investment returns for your clients. Nonetheless, getting your data architecture right as early as possible is a highly important differentiating factor. Without the necessary data, your business will not be able to offer additional value over your competition.
Data determines every aspect of the client relationship and the client's perception of you as an asset manager. From generating performance through better-informed investment decisions to providing clients with superior reports and digital channels. Finally, you want to avoid having to invest in expensive change projects on your data supply chain, every time you want to offer a new product or service. You will be able to optimize your maintenance costs, freeing up the budget for change projects that create direct value.
3. Data delivery and integration
At the beginning of the data supply chain, externally and internally generated data is ingested into a landing zone, a data lake, or directly into a structured data warehouse. This classic ETL (extract, transform, load), or data integration process is a major source of IT change costs and requires resources to ensure constant surveillance of timeliness and data quality. All processes need to be timed and checked for completion.
Here are some options and ways to tackle the above-mentioned issue
One-to-one sourcing, the "classic" route (e.g. via FTP/file, API)
In this setting, data is directly sourced from its provider, using a connector specific to the provider's specifications.
This somewhat antiquated method may still be required when using smaller or niche providers, you only need to connect to very few specific data vendors, or the involved systems do not allow other means of data transfer, for example for large data queries.
The use of a data integration layer can reduce IT effort and simplify data architecture
The integration layer provides one point to access data in different input formats and from different sources, decoupling the internal systems from the ETL process.
A universal connector from your integration layer to downstream systems can be used to reduce the number of internal interfaces.
If no data integration layer or provider is used, new connections require a significant amount of development, testing, adjusting of firewall settings, and integration into data management, and governance processes.
One-to-one connections may still be required to source real-time data such as instrument prices, as there are often specific to venues and display systems.
Use of data aggregators and cloud capabilities
A relatively new trend in the data landscape, financial data aggregators and exchanges, provide different types of data from a large variety of vendors using a single source.
Platforms like Crux, Open: Factset, BattleFin, Quandl, and Knoema cover a broad range of data sets from alternative data to index benchmarks, credit ratings, exchange pricing, and many additional sources.
Data can be received either directly from data exchange, or ordered through a cloud provider's data market place. For example, AWS and Snowflake offer financial data which is made available directly in the respective data warehouse, removing the need of building your ETL solution.
While data recipients often manage their subscriptions directly with the respective data vendor, some exchanges are also offering an e-commerce platform, where data can be licensed without contacting each data vendor directly.
Some data aggregators provide managed services such as delivery management through various supported connectors or formats, as well as validation of data quality before it reaches the client.
This form of data aggregation and distribution is still a nascent business, it is, therefore, necessary to check for potential gaps. For example, when looking for a specific benchmark index, its provision may not yet be available through a data exchange.
Use of a managed data service provider
As a separate option with the largest impact on the operating model, outsourcing providers offer to provide the entire data supply chain up to running the data warehouse.
The choice of data delivery method becomes entirely the responsibility of the outsourcing provider.
Suitability on the operating model, overall costs and offering scope, impact on data governance and data management processes need to be considered in detail.
The next important step is to ensure the definition and use of an enterprise data model, that will enable your organisation to leverage a single source of truth. For example, the same security or transaction should be defined consistently for all users, be it your risk team or your portfolio managers. You can build your own enterprise data model, or leverage existing ones for the industry, built and maintained by a service provider.
In general, we recommend to avoid connecting directly to individual data providers, unless there are no viable alternatives. Relying on a managed data service provider adds new oversight tasks to internal data management, but often relieves it from many manual or support related tasks. By careful consideration of the requirements, a scalable data architecture can be implemented, which will make a competitive difference in the market.
4. Data governance and data management
The data governance policy describes individual and team responsibilities towards data ownership, data access rights management, as well as responsibilities towards well defined data functions and data sets. A good data architecture is a prerequisite for the introduction of effective data governance, as it provides the necessary transparency and categorisation of data and related functions. At the same time, a good data governance is critical to the maintenance and consistency of a good data architecture.
Data management is a business function which serves as the central point of contact for all data related inquiries. It is the one team responsible for ensuring data quality, timeliness and its consistency across systems. The team works within the data governance framework, and takes ownership of specific data sets supplied to the different business functions.
An operating model which relies on in-house ETL systems and batch processing often comes with extensive involvement of data management into overseeing and interfering with the technical processing, as well as the need for an IT team to support them. Questions and issues related to data quality and timeliness from the different business areas frequently require extensive investigation and resolution processes.
The use of data aggregators or managed service providers changes the work description of the data management team. Usually, their responsibilities towards internal business functions will remain within the data governance framework. However, upstream responsibilities will be rather focused on an oversight function. Instead of having to interfere with processes directly, data management will be responsible to oversee the timely delivery of data from one external provider within a well defined SLA. Any IT related issues upstream from the asset manager's data warehouse are now in the responsibility of the provider.
Did this article provide you with food for thoughts? Are you facing data challenges, or thinking of improving your data architecture? Do you want a scalable and sound data architecture, enabling investment decision-making, high-quality reporting, digital distribution channels and business intelligence, enabling potential for innovation? Let us discuss how we can help you.