Analytics is the method of logical analysis, dating back to 1590 from Aristotle’s works on logic. Simplified, it’s the systematic analysis of data or statistics. While developing an analytics strategy may just sound like figuring out how to best analyze your organization’s data, it’s far more complex than that.
A comprehensive analytics strategy should not only determine how data is going to be analyzed, but should also address where your organization is today, where it wants to go, and how it is going to get there. It should articulate the long-term decisions needed around how data is going to be used, governed, and consumed to satisfy organizational goals and missions.
As straightforward as it may sound to create an analytics strategy, figuring out where to start can be tricky.
Data is often scattered in silos, trapped in legacy systems that don’t talk well with newer ones or data quality is fragmented through manual user processes. For all of the complexities that come along with developing a data-driven culture, it’s imperative that you have a foundational understanding of what is needed to create an analytics strategy.
In this guide, we’ll define what an analytics strategy really is, break down the steps to creating one, and give you some key tips to think about.
An analytics strategy is part of a comprehensive strategic vision to specify how data is collected and used to inform business decisions. It is meant to provide clarity on key reporting metrics by:
Oftentimes the terms analytics strategy and data strategy are used interchangeably, causing some confusion about what each one means. If we’re talking about data strategy, we’re focusing strictly on the characteristics of our organization’s data.
A strong data strategy should address things like data content, quality, ownership, and governance, as well as associated security and provisioning. Data strategy is the overarching operational strategy that encompasses both data engineering and data analytics.
Analytics strategy should translate the data strategy into an actionable plan to implement it. It should address organizational objectives, desired business outcomes from data, educate stakeholders, and establish a plan for implementing the strategy. With accessibility to data being greater than ever, it’s imperative that your organization has a plan to support accurate decision-making.
Ensuring an understanding of the current state of your organization’s data, processes, and data culture will help make sure solutions are sustainable and iterative. Analytics strategy will be the focus of this eBook.
Looking for more information on Data Strategy? Be sure to download our How to Build an Actionable Data Strategy Framework guide.
Now that we know that data analytics strategy is part of a comprehensive strategic vision to specify how data is used to inform business decisions and processes, let’s take a moment to discuss why it matters.
One of the most common phrases around building a data and analytics strategy is establishing a “single source of truth.”
For example, two departments, let’s say Sales and Finance, might both report the Profit metric back to the executive team. Lo and behold, their numbers are different.
Establishing a single source of truth includes the ability to specify the source, type, definition, and lineage of every data element. Knowing where the data originates and the changes that are made along the way is key. This allows the business to know and communicate what the data represents.
In the example we used here, maybe Finance is using Net Profit and Sales is using Gross Profit. This is an example of both a definition and lineage problem. Net and Gross Profits are calculated differently, something that is addressed by eliminating its lineage (i.e. how it’s calculated). It’s also a definition problem. Profit is too broad of a term.
Having more granular definitions of these terms (Net vs. Gross) removes ambiguity. Possessing several versions of data across an organization certainly isn’t a unique problem, but one that we aim to address with a comprehensive strategy.
Creating and communicating a “single source of truth” is a large portion of why we establish an analytics strategy, but not the entire story. It’s just a really good example. A comprehensive analytics strategy also allows us to:
To do these effectively, let’s review some steps we recommend our customers take along with examples and templates to help build a foundation for an actionable analytics strategy.
The first step of formulating an analytics strategy should be to identify your key players. These individuals should have a vested interest in the data platform, excitement about the use cases, and a long-term goal of making the organization more data-driven.
The group ideally should be cross-functional in nature to ensure that different interests of the organization are there to give input. Some examples of primary stakeholders include:
As you can see, these recommendations have representatives from a variety of business units, IT teams, and managers/executives. While the senior leadership of an analytical strategy may provide an important direction setting for the final reporting tools, many other players will likely contribute to how the data analytic strategy takes shape.
Don’t fall into the trap of leaders prescribing the vision without input from the business analysts or consumers. After all, these are the people who understand the data the best.
Ideally, someone within the data analytics team should run the key stakeholder engagements to ensure there is a perspective that brings an understanding of how the entire data platform architecture works together. Now that the team is established, let’s move to the next step.
PRO TIP: Don’t fall into the trap of leaders prescribing the vision without input from the business analysts or consumers. After all, these are the people who understand the data the best.
Ideally, someone within the data analytics team should run the key stakeholder engagements to ensure there is a perspective that brings an understanding of how the entire data platform architecture works together. Now that the team is established, let’s move to the next step.
The second step in developing an analytics strategy is to conduct several discovery sessions to uncover the current processes around data analysis in your organization. A comprehensive discovery session should do more than just establish primary use cases.
Initial sessions should cover the current state of data assets and platform technologies in addition to current use cases.
These discovery sessions can be facilitated in several different ways, whether with individual meetings with your stakeholder groups or even full group sessions. Your sessions should answer questions around the organization’s processes for centralizing, governing, and accessing data.
To do this effectively, you’ll need input from each group represented in your key stakeholders. Some common questions to ask during discovery include:
You’ll notice the first four questions are more geared towards your business users, while the second two focus on the systems that are going to get us what we need. The business users are going to help define and prioritize what should be worked on first in a perfectly accessible data environment, and answering the technical data platform questions will help determine feasibility.
Spend time assessing each data source and consider putting together an impact/complexity matrix to help target the low-hanging fruit. Be sure to include any relevant KPIs and their associated business value to help drive those strategic conversations with your executive sponsors.
Not sure if a KPI is relevant and feasible? Here’s an easy trick: Have stakeholders explain why a particular metric is important, but also ask to point out where the data that informs the metrics reside. If data is not available or is not consistent with the KPI metric, then there may need to be additional work, generally data development.
While gathering enough broad context is important, your primary use cases are going to drive most of these conversations. It’s how we connect the value to the effort of the entire strategy. The primary use cases should align with your business’s top priority and goals while also having the potential to be completely supercharged by data. Some considerations for your use cases are:
Analytical data products should inform decision-making and help analyze specific business functions. These use cases should have a significant impact on the overall business but understanding the current state of data cleanliness, accessibility, and data definitions will be critical to your use case’s success.
Make sure your use cases are valuable enough to justify the time spent, but not too complex that you lose executive sponsorship or buy-in along the way. Prioritize the use cases that hit that sweet spot between value and complexity, and continuously educate executives and stakeholders about what is possible and what to expect from your analytics initiatives.
Your analytics operating model, also sometimes called the delivery model, will inform many of the strategic decisions made around data. There are three common models:
Business Units and Functional Areas operate with complete autonomy, while attempting to maintain global standards to meet specific enterprise requirements.
Central team provides a single point of control at the enterprise level for decision making, as well as structures for the Business Unit level for decisions.
Central team provides a single point of control at the enterprise level for decision making, with Business Units & Functional Areas having little or no responsibility.
There is a lot to address with each of these models, too much for this blog post, but essentially it is delegated decision-making responsibilities around data. Do we centralize everything in IT or an analytics COE? Is every business unit responsible for its own data needs? Or is it a hybrid approach?
We generally recommend a Federated Model for most of our clients, meaning enterprise-decision making is in a centralized group but each business unit has autonomy for departmental-level analytics.
Make sure to have a model you are working towards. Your analytics strategy should reflect your current and aspirational models.
Now that we’re starting to hone in on some high-value use cases and our key stakeholders are bought in, it’s time to start thinking about what tools are going to be used for analytics consumption. Like any other IT investment, your analytics platform should be sustainable enough to support existing and future business requirements.
Since we have already identified the core objectives of our business units and desired outcomes of our analytics initiatives, this part should be relatively straightforward. Here are a few factors to consider when choosing a BI platform:
The technology assessment may reveal the need to do a proper proof of technology to build a better understanding of how the proposed tools stack up to the selection criteria. At phData, we have a strong background in implementing analytics platforms using a variety of technologies. The below selection criteria is what we use to help our customers maximize their investments:
Technology decisions are critical to a successful analytics program. The decision drives everything from costs to final analysis, and choosing a tool that doesn’t have buy-in across the board will prevent a successful analytics strategy.
Strike a balance between a tool that is easy to learn, provides the connection types necessary for the majority of any potential use cases, is equipped with the security standards needed, and your team is excited about using.
Your entire organization needs to understand the basics of working with data: how it is structured, how to do simple analyses, and how many of the manual processes they are completing could be automated.
They don’t need to know how to do it yet, they just need to be aware of how that could happen, and how valuable these skills can be to your organization as it potentially increases the capacity of each employee. Although it’s step 4 in our list here, establishing a data culture might be the most important and most difficult part of creating an analytics strategy.
Oftentimes, technical challenges can be overcome with more experts or with better tools, but cultural changes are much larger obstacles. A shifting mindset for ten, one hundred, or ten thousand employees can present a daunting challenge, but there are a few points to focus on to create a sustainable data culture.
In addition to all of the front-facing efforts in establishing data culture, make sure IT can support technology challenges. Waiting on access to data or struggling with licensing issues on software can all bottleneck analysis and create distrust in the analytics program. Ensure that IT is staffed and partnered with your analytics efforts, even if that means having a dedicated team outside of traditional operational IT.
Creating a plan to manage data requests and helping users understand data is our final piece to the analytics strategy puzzle. It is important to call out that implementing a data platform is not a simple process in itself, but it is something that will be more easily achieved with a solid data strategy. Much of this section will assume some level of data strategy has been considered and/or defined. The core capabilities of data in our analytics strategy should be:
With the development of centralized sources of truth, your data will ultimately run into some ever-present issues including changes in business logic and the need to bring together or harmonize disparate data sources. These requests will come early and often.
Data sources that were defined and developed change. Rules don’t stay the same, and it’s important not to meet these requests with a sense of frustration. Instead, it’s better to acknowledge that the requests are a sign of the process working, as business units are submitting requests to you and your teams.
As analytics programs and data platforms mature, the impact of not having specific components of a data governance program becomes more and more important. On the other hand, over-engineering a data governance program can slow down progress and limit business value.
The key is balance. Pay special attention to certain business units that have non-negotiable data governance requirements. Compliance regulations in certain industries require specific governance of the platform. Identify aspects of data governance that are non-negotiable and those that can be developed later in the data platform life cycle.
Strategy is an ever-evolving process. Just because you have created the 6 steps outlined above, doesn’t mean you are done. It will continue as a malleable ongoing task throughout your organization. To maintain an effective strategy, here are a couple of quick items to focus on to ensure sustainable success.
A successful strategy process should strike a balance between being rigid enough to provide guidance and direction, but nimble enough to course-correct as needed. This process needs to be owned internally by an analytic strategy product owner(s). This individual ensures the principles of the strategy are well promoted, reviewed, and followed.
This product owner may belong to an IT department or in a broader executive department that is responsible for coordinating metrics across the organization. Depending on the analytic maturity of the organization, a single person, unit, or entire division may be devoted to the overall analytic needs and organization used.
Initiatives in an organization don’t succeed, in the long term, unless there is responsibility and accountability. Make sure someone (or some group) owns the overall analytics vision and strategy of an organization. It is an evolving, ongoing, process after all. Somebody needs to keep the ship pointed in the right direction.
Any strategy development process will produce key documentation around the procedures, protocols, and outputs of interrelated processes. However, an analytics strategy engagement should also track the procedures, protocols, and outputs of the strategic engagement itself.
There will be a lot of documents.
We recommend storing all related documents on a wiki or shared drive. One, it’s great for referencing back after the fact. If your product owner leaves, you can have a new person pick up the initiative (relatively) easily. Two, you will need to consistently demo the analytics strategy. New hires, particularly senior leaders and executives, will need to be made aware of the analytics vision of a company.
Once an organization starts progressing on its analytics journey, you may then embark on determining what tools and technologies are best suited to reporting the progress made for its business analytics. Some tools and technologies are better suited for specific kinds of analytical reporting, while others focus on value for more generic reporting.
In the table above, four stages of analytic maturity are presented. Typically, more complex analytical processes bring greater value to the organization.
Descriptive Analytics – These analytical approaches typically answer the question “What happened?” In other words, they typically look at past data points and describe what took place.
Diagnostic Analytics – Presenting diagnostic metrics tend to be more explanatory than descriptive analytics. They generally answer “Why, or under what conditions and circumstances, did these results occur?”
Predictive Analytics – Many organizations aspire to achieve this level of insight from their reporting technologies. These analytics help build scenarios around “What will happen if…?” Typically, analytical frameworks at this stage require large quantities of data with many different outcomes.
Prescriptive Analytics – This stage is very difficult to attain, as it attempts to make a causal link between business activities and the desired outcomes. In many ways, this stage is similar to looking into the future based on current data.
Tableau and Power BI, for example, are great descriptive and diagnostic tools. However, they are immature predictive and prescriptive tools. As an organization becomes more analytically mature, you may need to invest in new tools. Dataiku, Alteryx, and Azure ML Studio may be better self-service analytics tools for predictive and prescriptive analysis.
Remember to evaluate how long the data analytic strategy should be used. Much like an organizational strategic plan, an analytic strategy should have a time-bound period of use before any kind of plan redetermination is required.
When considering the shelf-life of a strategic data and analytical strategy, several questions are necessary to be addressed:
Remember that different departments within an organization will be at different stages of analytics maturity. Different groups may need different tools to support their specific maturity.
Ideally, your analytics strategy will be living breathing principles that influence the way you approach your data challenges rather than set in stone commandments. It should be agile enough to promote individuals and collaboration rather than just focusing on the tools or technology.
It should be able to react to changes in scope or priority as new data becomes available or as business objectives change. Communicating early and often with all of your key stakeholders will ensure satisfaction and a successful outcome.
Our sincere hope is that this guide serves as a great resource to help guide you on your analytics journey as you evolve your organization. The general principles and best practices in this guide truly apply to organizations of all sizes, industries, and data/analytic maturities.
At phData, we know this work is foundational to your success and we are here to help. Drawing from years of experience, learning, iterating, and doing this for customers at every stage of their life cycle, we’ve built a set of tools, processes, reference architectures, and a team that is ready to help you get on the right path towards better-utilizing data and analytics within your organization(s).
This guide was expertly crafted by analytics experts Spencer Hamilton and Phil Perrin. Edits provided by Alex Christensen.
Subscribe to our newsletter
Data Coach is our premium analytics training program with one-on-one coaching from renowned experts.
