Design a Cloud Analytics Strategy for Success

Cloud solutions are often promoted as the panacea to all problems for both IT and end users. They claim to achieve user independence from IT in the form of self-service, freedom for IT from infrastructure maintenance, cost savings, and much more. There are elements of truth and yet much to take with a grain of salt, particularly around actual cost savings; support needs; and the level of access, control, and flexibility you have with cloud solutions. Cloud analytics can help you modernize your technical infrastructure, but as with any technology solution there are people and process issues that must also be addressed. ASR has worked extensively with dozens of clients over the past 15 years, and we are seeing this shift to the cloud first-hand. In this article we will share the benefits and key considerations for moving to cloud analytics so you can achieve the greatest results and truly make analytics your superhero.

Historical challenges with analytics solutions

By understanding the past, we can better understand how to avoid the mistakes that have held institutions back from making the most of their investments in analytics: a lack of understanding of the differences between operational and analytic technology, over-hyped vendor claims that create unrealistic expectations of analytics’ practical capabilities, the importance of campus-wide data integration, the need for ongoing data governance, and the difficulty of fully implementing an analytics solution without external support.

Whether built in-house or bought as an add-on module to an existing system, all components of analytics solutions differ significantly from traditional higher education enterprise resource planning (ERP) systems and stand-alone systems for recruiting, housing, advancement, etc. Those transactional systems collect and process data, are typically designed for the operational support of a single department, and are not optimized for data analysis across entities. Surfacing and transforming this data in an analytic application requires a different data structure and ultimately a different end-user interface, both of which put responsibility on the people and process aspects to ensure understanding and adoption by end users. Advanced analytic methods such as machine learning may require unstructured data not normally seen in transactional systems. Most of the traditional transaction system providers do not truly understand analytics and how it is different from their products’ core functions. Shoehorning existing infrastructure into bolt-on systems from third parties without really considering a comprehensive strategic approach does not a cloud platform make!

Effective analytics, particularly for advanced forecasting and analysis, requires integrating and organizing a broad spectrum of data from a range of source systems into a structure that provides time series data across the entire student lifecycle from prospects to alumni. Historical data is often more broadly defined than data used to establish point-to-point or enterprise integrations for the purpose of transaction processing. This is especially a challenge with derived or calculated values that have time frames around them. ERP providers typically do not have easy access to the data outside of their systems, leaving it to the client to integrate third-party systems for their analytic needs.

This fundamental disconnect between transactional and analytical systems is a major challenge to institutions attempting to build their own analytics environments without external, specialized expertise. Institutions often underestimate the effort and cost to build their own analytics system, which requires much more than technology purchases. The scope of an analytics implementation is substantial, with the infrastructure maintenance of databases, release updates, data structure change requests, among the possible issues. This can consume the limited and in-demand technical resources necessary for success. Changing user requirements and institutional priorities and staff turnover all impact the effort to maintain an analytics environment.

Institutions often lack the type of specialized knowledge internally to build, deploy, and maintain analytics because the needs of the college dictate broader skillsets when resources and positions are limited. Leaving the infrastructure development to the experts allows you to focus on your analytic goals rather than attempting to create the solution. There is significant effort and cost behind developing and implementing a cloud analytics solution, and there is no quick fix. When done properly, the returns are well worth the investment. The most common and immediate benefit is automating data integration and manual data manipulations to quickly view year-over-year trends. Automating this repeated manual effort adds up to hundreds, sometimes thousands of hours of effort saved annually.

Finally, most institutions struggle with establishing organizational structures for data governance that make an analytics platform effective across the diverse set of users, business processes, and needs. It takes a great deal of leadership to convince users that they no longer need their shadow data sets in Excel or Access, and departments that they need to use the enterprise analytic tool rather than purchasing their own software. This is particularly true in decentralized environments, where data silos and disconnects between departments already exist. In this regard, the move to cloud analytics is a culture shift as much as a technology shift. An analytics solution can provide data and insights, but you cannot ignore the fact it still relies on the human element to tell the story. Decision makers are responsible for applying their interpretations and experience to taking action.

Leverage cloud analytics' greatest strengths

Cloud analytics provides a promising architecture to address many of these historical issues. It can help you take advantage of the latest analytics features and capabilities, standardize data integration, apply consistent data definitions, and simplify training and the user experience, all while offloading much of the technical architecture and support tasks. This approach ensures you retain ownership of your data and improve user adoption. There are various paths to take on your journey to the cloud, each with different levels of commitment.

Like many other aspects of IT, institutions are not always experts in every possible technology. They may be better served using their resources to partner with someone who understands the unique design, support, and adaptation of the data structure and technologies required of a cloud analytics solution. The caveat is to not give up control of your data model or the ability to configure and adapt that data model and integrations to support your institution’s specific needs. That is not to say every cloud provider is open to every possible system, API, or unlimited configuration and tailoring. But having the appropriate level of access at this level is essential to success with cloud analytics.

Cloud analytics can help rejuvenate a stalled, incomplete, or inadequate deployment of an on-premise infrastructure, or to pilot in a business area lacking an effective analytic infrastructure. These techniques can be used as part of a long-term strategy to iteratively migrate into the cloud additional areas where the needs are greatest. For example, if your on-premise solution has a significant backlog of maintenance, the databases are on old unsupported versions, or the business intelligence platform is many releases behind, these situations are prime opportunities to rethink the analytics infrastructure and strategy and look to migrate to cloud analytics. This approach is especially useful when trying to crack the mystery of a particularly challenging business problem. You may not have the scope and depth of data necessary to effectively apply analytics to come up with a reliable and applicable model. In this case, you use the necessary cloud analytic technologies as a service without having to put forward as significant an investment in internal technical infrastructure. If the analytic results are incomplete, inaccurate, and unusable (which happens more often than many institutions are willing to admit given the small data sets common in higher education) then you simply discontinue use and subscription of the analytic infrastructure. Think of it as a “try and buy” approach.

Another driver for cloud analytics can be piggybacking off another major cloud initiative. Perhaps there is a plan to move the LMS to the cloud or migrate systems to the cloud as in-house hardware is reaching the end of serviceable life. This reduces risk, keeps attention focused, and provides a purpose to the project, all of which are necessary for success.

Cloud solutions have better optimized integration of databases and configuration for specific analytic needs. For example, there are new cloud tools that automatically perform the necessary change data capture to allow you to “time travel”, creating the data structures necessary to see trends and patterns that drive decision making and predictive models. There are plenty of defined integrations to existing data sources and APIs that allow you to leverage the source data systems from your environment in your specific combination. This architecture does not have to be driven exclusively by the main ERP system provider or the companies and systems they have partnered with. You have access to powerful cloud tools for data modeling and structure that let you define data the way you want to, either internally if you have the expertise, or using an external partner.

Cloud analytics provides a great platform for staying on the latest supported versions of everything and having access to the latest features, integrations, and data and API structures. This is important for adapting to changing requirements. The cloud environment by nature is resilient and supports effective high availability, redundancy, and disaster recovery. Another benefit is the growing specialization of the database design and processing seen in cloud platforms that embed common services in a tightly coupled fashion, allowing you to leverage the optimized design and infrastructure built specifically for the analytic purposes.

Any cloud approach should address this fundamental goal: to make the analytic solutions work for the user and support the institutional strategy. To achieve this, it’s important to find a platform and vendor that takes a unique approach compared to most cloud or software as a service (SaaS) applications. Most SaaS offerings restrict the ability to tailor or manage your own data definitions, workflows, and visualizations. You need the cloud analytics platform to give you the flexibility to represent your business needs, not a rigidly defined design compromise to support complex requirements in an industry as diverse as higher education. Several of our clients have experienced the limitations of SaaS when moving their ERPs to the cloud, largely due to loss of control over data definitions and the timing of upgrades and customizations. However, this need not be the case for cloud analytics. It is essential to be able to define data that reflects your strategic and operational goals, pull data from your set of systems, and organize it all in a way that reflects how you do business. Further, while it is helpful to have a template of defined data structure and analytic models and visualizations to present to users, one size does not fit all. A good cloud analytics platform and strategy gives you access to tailor the data model to your needs and systems environment while allowing you to present data in a way that informs decisions and spurs action. This is an essential requirement and knocks down one of the greatest barriers to institutional success with analytics.

Cloud analytics still needs effective data governance

While the use of cloud analytics can improve your institution’s decision-making and ultimately improve student outcomes, the issues related to people, process, or budget should be given as much attention as the technology. Careful consideration should be given to these areas to ensure that leadership and users are asking the right analytic questions, and uncovering and improving business processes that create data quality issues, and using the insights to drive action.

Even superheroes have their kryptonite, and for cloud analytics it is the need for data governance. A truly successful cloud implementation requires strong processes and buy-in from people at all levels of the institution. The project team must have the authority to address these needs and put basic parameters in place, with full support from leadership. Though most cloud tools improve the way we collaborate, share, and communicate, these tools are ultimately used by people. Whether your analytics solution is partly or fully in the cloud, the most important part is what people are doing with that data and how they are processing, interpreting, and presenting it.

Data governance provides an organizational structure to analyze the root cause of data quality issues, identify and implement improved business processes and training for best practices, and operationally define the key metrics that are critical to your institution’s analytic goals. This addresses the critical factors of understanding your data and data needs and developing the ability to tell data stories effectively. While analytic tools can help identify issues in the data, they cannot fix them. Successfully leveraging analytics depends on selecting a competent, proven, and trusted analytics infrastructure provider who knows your business. By leveraging their skills and experts who are focused on that job 100% of the time allows you to redirect internal resources to the areas in which your institution excels – understanding your people, processes, and students.

Without careful monitoring and management, cloud solutions are not necessarily cheaper than on-premise solutions; this is a common misconception. While cost may be spread out over longer time and appear cheaper at the beginning, offloading the support and design of the platform to a cloud provider also offloads those costs to them, costs they will seek to recoup through implementation and support fees. You need to research carefully what support and services are included and what is not, what is fixed in the subscription fees and what gets added on top or added after maximum usage thresholds. It can be very difficult to estimate or predict the variable costs in cloud infrastructure for data storage, data transfer, or excess CPU/workloads. You must understand and use the monitoring tools provided and how they relate to usage to avoid unexpected overruns of monthly costs. Cloud storage is not unlimited, certainly not without paying for it. Most cloud analytics services are metered in some way even if the cost appears fixed, so it is possible to have surprises. Finally, even a fully hosted cloud analytic platform does not eliminate all the work from internal IT staff. Their workload may lessen, but it will also shift to managing the integrations to cloud and relationships with the cloud partners.

How cloud analytics can be your superhero

There is justifiable skepticism from those who have worked in higher ed technology for a long time and seen many cycles of analytics hype and promise go partially or completely unfulfilled. It is fair to be cautious given the tendency in the technology industry to repackage the same old solutions with a new name, but not have really solved many of the fundamental problems that existed in the preceding solutions. Part of this is driven by the interest in taking advantage of new technologies and approaches without having truly heard the voice of the customer or understood the fundamental root causes of the problems that business users face.

By understanding the areas where cloud analytics can help, and the areas that still require the human touch of people and process, you will be best positioned for success and achieving the benefits of the flexibility and power of cloud analytics. Consider the following approaches and benefits in your cloud analytics strategy:

  • Use cloud analytics to pilot a specific area of inquiry and identify the institutional strengths, weaknesses, and challenges deploying in a new architecture.
  • Use the cloud to progressively upgrade and migrate your current analytics environment. Deprecate what is not working, migrate what is, and enhance the analytics infrastructure where capabilities are missing.
  • Cloud is a great way to start fresh if beginning your analytics journey and trying to move beyond operational reporting from transactional source systems.
  • Whether migrating an existing infrastructure or starting new, using cloud analytics in a carefully planned, staged, way allows you to move without putting all your eggs in one basket. If issues arise or requirements change, it is easier to pivot and consider alternatives before the cloud analytics environment is complete, unlike most on-premise approaches where investments in hardware and software are typically committed early on.
  • The institution has less specialized infrastructure to worry about and manage. Even for larger institutions, it is worth considering if this is the best use of IT’s time and skills. Consider what tasks and knowledge adds the most value in analytics for the institution. It’s not the technology itself – it’s removing the repetitive automated data integration and maintenance and letting people focus on the use of the data applied to business problems that serve students, save money, or impact growth and efficiency.
  • The cloud natively supports dynamic processing needs and spikes in demand typical of most analytic processing. The institution does not have to size for peak loads that go wasted most of the time or develop their own virtualized infrastructure to distribute workload where and when it is needed.
  • Cloud infrastructure can improve the access and delivery of visualizations, interactivity, and data understanding and sharing with users, particularly in remote and distributed work environments.
  • The cloud still provides access to the analytic models and logic for your own adaptation and understanding. You are not necessarily giving up the access to and understanding of what analytics are doing. (There are several "black box" systems like that, but we do not recommend this approach for transparency and ethical reasons. To understand the risks in that approach, read the story of Theranos in the book Bad Blood. Hiding behind "proprietary models" and "patented systems" led to one of the greatest financial and technical frauds in recent history.) We always recommend maintaining control over your data definitions and access to and understanding of what the modeling/processing/logic is doing.

The COVID-19 pandemic laid bare the need for easy access to a wide range of understandable and usable information for planning and decision making. Many colleges found themselves lacking in this area when they needed it most. The gaps in data collection, overlapping analytic tools, disconnect departmental systems, and outdated analytic technology infrastructure point to the urgent need to modernize.


About ASR

ASR has been providing analytics solutions to higher education for over 15 years. Our approach is different from other vendors - we are real-world implementers that have one fundamental goal: to make the solutions work for clients and help them achieve the institutional and student outcomes as they define success. Whether the client chooses completely customized implementations or our tailorable Student Success Analytics (SSA) solution, we engage with them to tailor analytics to meet their specific needs, business processes, and data profile.