7 Easy Steps to Building an Actionable Data Strategy

What is a data strategy (and what is not)?

A data strategy is a plan defining long-term data objectives, as well as the data architecture, tools, talent, and short-term initiatives needed to achieve them. 

But don’t mistake a data strategy for your current data tasks. The most recent Jira tickets on a miscalculated metric or failing data integration job are not part of the data strategy.

Unlike traditional data management and project management tasks, a data strategy:

1. Defines the goals of the data beyond quarterly achievements.
2. Is focused on data-driven transformation - what can be achieved to change and elevate the current use of data across all business processes and raise the value of the company’s data across the entire organization.

To better understand how a data strategy looks in practice, let’s look at its key features.

What are the key components of a data strategy?

No two enterprises will have the same data strategy. 

But every good data strategy will include the following key components:

1. Alignment with the corporate strategy, with an explanation of how the data strategy helps achieve long-term business goals.
2. A clear prioritization framework for which data initiatives are going into deployment (and which will be avoided). 
3. The data architecture with an overview of technologies, platforms, the current data assets, and their lineage throughout the data lifecycle.
4. The team who will be responsible for the execution and oversight of the data strategy. 
5. The data governance framework outlines standards for data access, security, and data quality.
6. A clear roadmap specifying the milestones and timelines for the implementation of the data strategy.

A successful data strategy will benefit corporations in multiple ways.

Why is it important to have a data strategy?

The right data strategy helps your organization to:

1. Make better decisions. Without high-level guidance, data operations quickly deteriorate to serve urgent requests instead of impactful initiatives. Consequently, changes that would better business models, data-driven decision-making, and elevate the value of data are put aside for the resolution of immediate business needs. 
2. Resolve tradeoffs faster. Without a data strategy to prioritize and align all data operations, each business unit will fight for dominance. From pressures to prioritize data collection of one department over another to definition wars over metrics and which business unit’s formula should prevail. A data strategy sets up clear priorities that disambiguate these tradeoffs and helps you streamline your work without arguing over “who should be next.”
3. Break down silos. Left by itself, each business unit will come up with metrics and definitions that serve their department, not the company as a whole. A good data strategy will focus on unifying the different perspectives on data, standardizing metrics, and providing the data assets for every department that break down silos and align work across disparate groups.

Let’s dive deep into the framework of how to build and deploy a successful data strategy.

7 steps to build an actionable data strategy

The following framework outlines the 7 crucial steps for building a good data strategy:

1. Identify the main purpose of your data
2. Perform a gap analysis
3. Prepare a proposal and get stakeholder buy-in
4. Build a team
5. Pick the right data architecture and tools
6. Set up data governance
7. Implement the data strategy roadmap

Step #1: Identify the main purpose of your data 

You should align the main purpose of your data and data strategy with your company’s long-term business goals and business strategy.

Having a clear prioritization framework will help you resolve conflicts about what should be implemented first and give you leverage to say no to data initiatives that do not contribute to your data strategy goals.

For example, imagine you are approached by both the VP of Sales and the VP of Logistics with their requests. The Sales leader wants the data team to automate data collection from CRMs and other sales data sources so they finally understand the prospects they’re working with, while the VP of Logistics asks you to clean the metadata in their data sets to remove confidential information when sharing data with supply chain partners.

Both are valid requests. Both can further business goals. But both require a lot of resources. Having a clear data purpose will help you navigate this prioritization dilemma.

If you’re uncertain about how to translate the business strategy into data purpose, tap into the two types of data strategies by Harvard Business Review: offensive vs defensive data strategies.

Defensive data strategies focus on minimizing downside risks - ensuring regulatory compliance, using analytics to detect and limit fraud.

Case study: Mall Group’s defensive strategy

Mall Group, the leading e-commerce group in Central Europe, used Keboola during a massive company expansion with the purpose to consolidate and streamline their operations. They purchased and integrated over 30 different companies servicing over 5M customers within 24 months. With Keboola, the team managed to not only integrate the unimaginable complexity of 30 different companies during the mergers and acquisitions. But also set up impeccable data security, regulatory compliance, and data governance throughout their data operations. Check the interview with Mall Group’s Head of Business Intelligence for details.

On the other hand, offensive data strategies focus on achieving business objectives such as increasing revenue, profitability, and improving customer experience.

Case study: Česká spořitelna’s offensive strategy

Česká spořitelna is the biggest Czech retail bank. Traditionally financial institutions focus on defensive strategies. But Česká spořitelna wanted to gain a competitive edge. So they used Keboola to build offensive growth engines such as optimized lead generation across the omnichannel, hyper-personalized marketing at scale, and automated credit risk scoring. Check the full story here. 

All companies will have to strike a balance between defensive and offensive approaches. But understanding which one your data strategy prioritizes helps you build a prioritization foundation.

Step #2: Perform a gap analysis

A gap analysis helps you identify what technologies, talent, or processes you’re missing to elevate the current state into the desired state.

To better understand your goal state, rely on Gartner’s data maturity model which specifies how organizations progress through different stages of data growth: descriptive, diagnostic, predictive, and prescriptive.

Not all company units will be at the same level of data maturity. Some departments will use big data predictive algorithms in their work, while others will struggle to track basic performance metrics. 

The general philosophy is to identify the lowest level of the data maturity model throughout the company, fortify it, and then move to the next stage.

Step #3: Prepare a proposal and get stakeholder buy-in

Once you’ve established your main purpose (step 1) and identified what needs to be changed (step 2), you can prepare a proposal for implementing your data strategy.

Your proposal will outline all the key components: 

• Alignment with the business strategy
• Prioritization framework
• Data architecture
• People
• Data governance
• Provisional roadmap

Executive sponsorship will be crucial for the success of your data strategy.

To make your case for change more appealing, focus on the return on investment gained by implementing your data strategy. For defensive data strategies, the value of data lies in costs saved and avoided. For the offensive data strategies, the value of data is derived from the additional revenue and profits.

Step #4: Build a team

To successfully implement your data strategy, you will have to build a team around your initiatives. But not every team member will be involved in the day-to-day operations. The right team will include:

1. Data professionals. These will be your day-to-day executors, who will drive the strategy forward by building data pipelines, architecting new solutions, analyzing data, or developing data science models.
2. Business leaders. Heads of departments will help you provide legitimacy and goodwill to implement the cultural changes for the data transformations. 
3. Data consumers. Regular check-ins with the beneficiary of your data strategy will give you feedback on how the transformation is improving business processes on the frontline.
4. (Optionally) data stewards. They help translate between the business and data/technical aspects of your data strategy. Data stewards are especially important in technically challenging operations and companies with complex data assets, where data literacy is not high enough for business users to self-serve their data needs. 

Understanding what headcount is needed helps you plan the hiring and training initiatives to get the right team. For example, maybe your in-house data analysts need more training in data analytics to provide the visualizations for descriptive analytics. Or you might need to hire more data engineers to optimize the data pipelines used by data scientists to build a predictive model.

Best practice: A lot of the issues with data literacy can be solved with a data catalog. A data catalog documents every field in your data set, allowing people to more easily self-serve their data needs and understand the data assets they’re working with. 

Keboola offers a Data Catalog that helps you go beyond documenting data. Keboola’s Data Catalog also binds itself to data, to be always aligned with data changes, and can easily be shared whenever you share data. 

Schedule a call to learn more about it. 

Step #5: Pick the right data architecture and tools

The data architecture will reflect the goals you’re trying to achieve. 

Data architecture decisions cover anything from:

1. How to build a scalable infrastructure?
2. How will users access the data assets? With separate decisions for technical and non-technical users.
3. What are the total costs of the data architecture?
4. Which specific tools and software solutions will we adopt?

For example, let’s say you’re looking for a data storage solution:

1. If you want to establish a single source of truth and break down silos, you will deploy a data warehouse with centralized and validated data assets such as metrics.
2. If you operate with huge big data datasets, you will choose a data lake to store the vast amount of unstructured data.
3. if, on the other hand, you need new data to come in in real-time, neither the data lake nor the data warehouse will be an ideal data architecture choice. Instead, you will opt for a data streamlining tool, such as Kafka.

Best practice: Picking the right tool for the job can save you time and money. For example, Hari built an artificial intelligence product with a 3x smaller team by having Keboola take care of every data process: from collecting data from various data sources, validating it, building predictive models, and finally embedding the machine learning algorithms into the product. 

Step #6: Set up data governance

Data governance will sit at the intersection of your processes and data architecture. The data governance framework defines:

1. Data access - who can view, manipulate, and share data. 
2. Data quality - the minimal standards for data before it can be exposed to data consumers.
3. Data security - policies that will keep data safe. Everything from encryption to sharing permissions.

Data governance will sometimes conflict with data architecture. Within data architecture, there is a trend to choose the best tool for the job, thus increasing the number of specialized solutions for their use cases. More tools increase the difficulty of applying data governance across all those tools. So you need to make a tradeoff between increasing your tooling surface and keeping data managed. 

Best practice: Governance and data architecture can both scale together with a common underlying infrastructure. For example, Keboola allows you to integrate multiple tools without increasing the burden on data governance. Keboola acts as a centralized command center through which you can set up governance rules for every process in the platform. All tooling can be tracked by following their lineage in Keboola. And security is guaranteed for all data.

Step #7: Implement the data strategy roadmap 

Finally, you start implementing the roadmap for your successful data strategy. The roadmap will specify the milestones and timeline to achieve them.

Make sure to include some quick wins in your milestones. Those will act as motivators for your team and signal success to the business leaders supporting your initiatives.

Best practice: Don’t be afraid to change your data strategy when needed. Re-evaluate your progress regularly, and if the data strategy becomes unaligned with the business strategy, make sure to course-correct.

Streamline your data strategy deployment with Keboola

Keboola is the data platform as a service that automates the data processes supporting your data strategy:

• Build your desired data architecture in a couple of clicks. Keboola’s plug-and-play design allows you to quickly integrate all the data tools you wish, while Keboola takes care of the underlying infrastructure and DevOps.
• Data professionals can work alongside business users within the same platform, by relying on low-code data tools and no-code features for the non-coders.
• Guaranteed best-in-class security standards out of the box. 

Keboola helped many companies establish or revamp their data strategies at a fraction of the costs. 

Wondering how Keboola can help you scale your operations into a better data future? 

Get in touch and let’s start planning your data transformation together.