Data Strategy for Executives

No, you aren’t doing enough with data.

Data is crossing all industry borders and continues to grow at a rapid pace. With this flood of information comes opportunity – but only for those willing and able to harness it.

It’s not enough to silo data to your CTO or IT team – for organisations to thrive in this age of information, a strong data strategy needs to be a priority of executives across the organisation.

What is a data strategy, and why is it important?

The What

Strategy is simply the interplay between decision making and timing. 

A data strategy limits the scope to just decisions regarding value-creation from data. If implemented correctly, your data strategy should positively impact revenue, customer acquisition, learning & development, resourcing, IT, and operations (any much more) while addressing the specific pain points of each department. 

The Why

Business is a kind of game with various teams, players and rules. Everyone engaged as either an employee or employer is playing the game on some level. 

As an executive, there are plenty of moves you can make. You can allocate capital to marketing campaigns, engage in strategic partnerships, hire staff to work on interesting problems, or even change the geographic location of the organisation. The possibilities are endless!

Here’s the problem – much like the university student who is told they can “become anything” yet struggles to decide what to major in, a greater number of decisions increases the complexity of determining what is optimal. 

When you add in the element of timing, continual perfect decision-making from any single entity is close to impossible. Engaging in an infrastructure project might be a good decision during the start of the financial year while the coffers are full, but may be a mistake when cash flow is tight. 

This is where data comes in. Creating a data strategy is the ultimate key to improving all future strategies. Strategy-ception, for the Christopher Nolan fans.

Ultimately, the goal of becoming data-driven is to mitigate the likelihood of bad decisions being made, and open up opportunities for new, innovative decisions. Data provides decision-makers with greater organisational awareness, safeguarding them against incorrect decisions that seem correct with partial information. Sending flowers is a lovely gesture – unless the recipient is allergic.

Informed staff make better moves, but informed executives ensure that we are playing the right games. 

How to create a data strategy

As you begin to embark on your data strategy journey, be aware that there is no “one size fits all” approach. A helpful first step is to define data use cases for your business context.

As previously mentioned, data transcends silos in your organisation. Ensure you don’t pigeon-hole your team to only think about traditional IT functions. What pain points do your marketing team members encounter? What about human resources? Operations? 

As QuantSpark so succinctly put it in their whitepaper titled Introduction to Analytics for Business Leaders, “just as oil produces everything from petrol to plastic, toothpaste to tires, the applications of data analytics are endless.” Your level of both domain and data knowledge is the eternal bottleneck for determining the correct data strategy – addressing this through seeking credible information from trusted sources will arm you with an awareness of what opportunities actually exist.

Personnel & capabilities

When it comes to data, capable – and technically proficient – individuals are essential. The technology to solve the majority of general data use cases is readily available – possessing the right personnel and governing structure to utilise the technology effectively is the hard part.

Assessing what data capabilities exist within your organisation is a good first step. This will lay the foundation to determine what is actually realistic. Naturally, any new data-specific staff you contract or hire are going to influence this greatly.

The next step is determining what data initiatives are aligned with your overall business strategy, and would therefore be impactful in the long-term. If you break this down further and determine the smallest amount of work needed to obtain a positive ROI, you can begin to cap the downside. Data initiatives shouldn’t be binary – a “failed” project should still produce real business value along the way. If a data project has absolutely no use until it’s finalised, you are unnecessarily increasing its inherent risk.

After you know what you want to do, it becomes a project management problem. Comparing bespoke vs out-of-the-box solutions, analysing system requirements, prototyping dashboards, and examining potential data models all come into play here. 

Incorporating data into the business strategy

Your data strategy should be intertwined with the overall business strategy. Consider what problems you are repeatedly coming up against, and assess what information would be needed to solve the problem. The best way to do this is by running a strategy day with all key decision makers, led by an agile delivery lead. Really drill down on your organisation’s strengths and weaknesses, and be willing to align yourself with the reality of the situation. What are your immediate concerns? What about further down the track? Are these problems industry-wide, or applicable just to you?

Another way to approach it is to consider what kinds of past decisions have gone badly due to incorrect assumptions, or even what decisions you have been unable to make due to uncertainty. If your leadership team is competent, most of your assumptions are probably correct – but in a way that makes them more dangerous. Start considering which assumptions you rely on regularly, and the type of data that would be needed to test them.

At the risk of sounding like a sales guy, at Aginic we literally do this for a living. Even if you aren’t after paid advisory right now, feel free to shoot us an email – happy to offer some pointers. I’d love to put together a post that perfectly caters to any and every scenario, but the reality is that the exact strategy is going to depend on your specific goals, personnel, infrastructure and business model.

That said, there are definitely some foundational principles that will take you pretty far on your own. We will cover these in the next section. 

Foundations of data strategy

Data must follow a fairly regimented journey to actually become usable. This is commonly referred to as ETL (Extract, Transform, Load), and there are dozens of resources that can cover that process.

What is less spoken about are the commonalities that underpin every facet of the data transformation process. Regardless of individual circumstances, the foundational components you will need to address include;

• Governance (How do I ensure my data is of a high enough quality to use?)
• Security (How do ensure the data doesn’t fall into the wrong hands?)
• Access (How do I ensure the data makes it into the right hands?)
• Responsibility (Who is ultimately responsible for which areas?)

Each of the above concepts are interconnected, meaning that failure to address any one of them may adversely effect the others.

Governance

Do you trust your own data?

Data governance is the set of policies and procedures in place to ensure your data’s security, integrity, availability and usability. Incorporating data governance into your data strategy is imperative, as it ensures that the data being utilised is of a high enough calibre to warrant trust. At its heart, implementing effective data governance requires three key things: a comprehensive understanding of the use cases, requirements and risks; sufficient technical skills in the implementation tool; and a willingness to invest time and resources up front, to enable scalability and flexibility down the track.

Data governance, if implemented well, helps ensure your business remains reputable and provides value. It’s also often needed to abide by laws, audits and regulations that can vary by location, industry, legislation, the size of your organisation and more. Fortunately, building a cohesive system for protecting data quality is highly achievable through implementing data quality principles.

Security

One security breach or a data leak can permanently damage an organisation’s reputation. There is no room for error when it comes to private data, so observing best practices is essential. The weakest link in security and privacy is always the human element, so regularly conducting security training for all staff is a great routine to get into.

The concept of data sensitivity is important here. Firstly, determine whether the sensitive data actually needs to be collected. If it doesn’t, remove it immediately and document your reasoning. If it does, see if you can strip identifiable information in some way and anonymise the data. For all pipelines and storage systems that still contain sensitive data, red team detailed attack and leak scenarios – a healthy pinch of paranoia is arguably beneficial.

Access

What is the point of constructing beautiful data cathedrals if the right staff members can’t easily access the data?

In conjunction with the last foundation element, applying the principle of least privilege is useful here. Don’t just hand out blanket admin privilege – allow individuals access to only the information they need to do their task (the further down the pipeline the better), and reduce the friction for granting permissions. The ticket system for obtaining access needs to be efficient, or the lag time between request and insight will negatively impact usage. An initial burst of effort here will keep access quick and safe for the future.

After establishing who needs what, the most important variable to consider is retrieval frequency. We can think of this as the data temperature – “hot data” is data that is queried often, while “cold data” sits in storage without human interaction for extended periods of time (as a writer, I can relate). Different tiers of cloud storage can help efficiently store data based on retrieval, and should be implemented accordingly.

Responsibility

When data projects are in their infancy, there is the tendency for the lines of responsibility to be blurred. We recommend deliberately taking time to codify roles and responsibilities around data domains early on, as the need for clarity will only increase as the data grows. Responsibility can occur at the table or log stream level for smaller operations, or even at the field entity level for more established businesses. 

Implementing your data strategy

Data strategy is not an assignment you finish – it will evolve with your business.Think long term, but prioritise which use cases are most urgent to address, before you focus on widening the scope.

Keep an agile mindset and strive for continuous learning and improvement as you iteratively implement your data strategy. What worked well and what can be improved as you scale your strategy? 

The progression of an organisation towards becoming data driven occurs in three (3) stages;

1. Cognitive
2. Associative
3. Autonomous

Depending on how advanced the capabilities of your organisation are, your answer to “what should I do right now” will differ. The stages are as follows;

Stage 1: Cognitive

Common company traits

• Data architecture and infrastructure are in the very early stages of planning and development.
• Virtually no decisions are driven entirely by data, and the ones that rely only on financial statement data.
• Few, if any, roles designated.
• Any requests for data are ad hoc, and the process for doing so is not widely known or formalised.

What you should do

Get buy-in from stakeholders – particularly other executives. Unless the leadership team sees the potential value from a project, it is going to be hard to get the multidisciplinary support needed to see it through.

Start mapping out what your current data architecture is, and compare this to what you are aiming for. This means determining business goals and the competitive advantage you’re trying to achieve with your data initiative. You will likely need to really get into the weeds of your data, and understand limitations.

Your first project should be a thin, vertical slice. Try to isolate a single area that is crying out for a more objective basis for decision making. This might be a single dashboard for a subsect of the marketing team, with the corresponding data architecture feeding into it. Executing a small project well will give you a taste of the challenges of a larger project, while also boosting confidence of stakeholders in your ability to deliver. 

Potential traps

There will likely be more conservative members in your organisation that may oppose the data project. Failure to produce visible “quick wins” early on will mentally confirm their doubts, and will harm overall organisational motivation for the initiative. Small, safe wins first, big wins after.

To assist with this, don’t box yourself in with unnecessary technical complexity. Use off-the-shelf solutions wherever possible, relying on their completeness to carry you safely over obstacles that you are unaware of. Custom solutions certainly have their place, but only when infrastructure can support the added workload through to the point where the custom code actually creates a competitive advantage.

Additionally, involve all stakeholders as early as possible – the wrong problem solved well is useless. Working within an agile framework will take care of this issue naturally, and is our strong recommendation for all ambitious technical projects.

Stage 2: Associative

Common company traits

• Formal data practices exist, with various end-to-end pipelines in existence throughout the organisation.
• Responsibility for the data is both centralised and distributed, with executive sponsorship and diversified users/stakeholders.
• Introducing new data sources is possible with some difficulty.
• Decisions are being influenced by the new information being generated.

What you should do

Begin to establish formal data practices. Adopting DevOps and DataOps principles throughout the organisation will pay dividends as the sheer quantity of your data exponentially grows. At this point, the data architecture can be modified to become increasingly more scalable and robust, with the nuances of your particular business model being incorporated into the overall design.

Start to explore opportunities to implement data science as a final layer on top of a refined data landscape. Upskilling staff is essential here – internal data specialists should be making deliberate efforts to develop deep domain expertise, and leadership teams should regularly discuss possible implementations of more advanced analysis for improving decision making.

Potential traps

There’s a temptation to adopt whatever new technologies make it on to a Thoughtworks tech radar. Avoid doing so unless it is clear that the value you create is worth the implementation cost.

Stage 3: Autonomous

Common company traits

• The ability to conduct self-service analytics exists throughout the organisation, however most data requirements are covered by formalised reports and dashboards that are an integral part of the company ecosystem.
• Introducing new data sources is straightforward, and current processes are well-documented.
• Data roles begin to specialise, with clear delineation between responsibilities. Full-stack data all-rounders have been replaced with specialised teams of engineers, analysts and scientists.
• Proper controls and practices are in place, and DataOps best practices are consciously practised.

What you should do

Create automations, particularly for the seamless introduction and usage of new data. Try to find ways to leverage data engineering labour hours for increasingly more impactful results. Audit inefficiencies ruthlessly, and structure trial phases for new automations.

Focus on community and communication. Follow the Bridgewater model of cultivating a radically transparent organisation where actionable data is disseminated throughout. Encourage collaboration, and allow staff to speak openly no matter their role. Pain points should be shared and collated in a single source, with the goal of continual innovative solutions for solving them using data.

Potential traps

Avoid becoming complacent. The main bottleneck should, and likely will be, the capacity of the data engineering team. Assuming that you can and should take on every data initiative under the sun will increase your exposure. Assuming you have already reached the pinnacle of being data-driven will hamper progress. Balance risk against business value, particularly as the labour needed for implementation increases.

Why isn’t every organisation data driven?

Data is hard. Most data projects fail.

As with any disruptive innovation, there are a large number of implementers that attempt to capture the significant value potential of the offering without fully understanding the underlying mechanisms of its value creation.

Data is no different. Without having expertise and collective organisational capability in both data engineering and analytics, ensuring that the particular nuances of your business are covered in a comprehensive vision is a colossal task. 

Here’s the pitch (skip this if you aren’t interested in paid data solutions) – our entire organisation is built around the idea of injecting highly talented data specialists into client companies to tackle their unique problems head on. We have helped countless happy clients execute ambitious data projects, and are growing rapidly. By providing organisations with access to our pool of top-tier engineers, we give them the means to put ideas into motion.

Whether you choose to rely on internal talent acquisition and upskilling, or prefer to bring in a specialist data consultancy like Aginic, experienced professionals are the real data asset. Remember this always. 

Not sure where to start? That’s okay! There is a lot to take in. Reach out today for a no-obligation chat, and we will help point you in the right direction.

Good luck on your journey to becoming data driven!