paint-brush
Your Data Strategy is a Cracked Water Bottle in the Desert (and How to Fix It)by@liorb
202 reads

Your Data Strategy is a Cracked Water Bottle in the Desert (and How to Fix It)

by Lior BarakApril 18th, 2024
Read on Terminal Reader
Read this story w/o Javascript
tldt arrow

Too Long; Didn't Read

Data strategy isn't about the bells and whistles; it's about clear, actionable goals. Before diving into infrastructure or systems, define the 3-5 key performance indicators (KPIs) These are your North Stars, guiding everyone toward what matters most, and are the top of your priority to secure.
featured image - Your Data Strategy is a Cracked Water Bottle in the Desert (and How to Fix It)
Lior Barak HackerNoon profile picture


Edmund Hillary, a New Zealand mountaineer, once said, “It is not the mountain we conquer, but ourselves.” In the scorching heat of the business landscape, your company is a thirsty caravan desperate for insights to guide its journey. But beware, for many, their data strategy resembles a cracked water bottle in the desert – leaking potential and leading nowhere. Are you ready to quench your company's thirst for success with those three simple steps?


Many leaders, desperate for a solution, picture a grand data strategy – a sprawling oasis filled with complex architecture and expensive tools. They gather their "data people," spend heavily, and build this technical utopia. But here's the surprising truth: most companies never reach that oasis.


Why? Because data strategy isn't about the bells and whistles. It's about clear, actionable goals.


Set Clear KPIs!

Start with the Destination, Not the Journey

Think of your data strategy as a roadmap. Before diving into infrastructure or systems, define the 3-5 key performance indicators (KPIs) that truly measure your company's health. These are your North Stars, guiding everyone toward what matters most, and are the top of your priority to secure.


In my book “Data is Like a Plate of Hummus’ which is based on my experience as a data leader and consultant, I many times refer to the KPIs refinement process as a funnel structure, keep things simple, keep them small, just like hummus, a very simple dish but with such powers to fill you stomach for the entire day, in my book I describe my method of refining the KPIs by starting with a set of questions, then pick the most relevant questions, and define which KPIs can answer them, the next step is to identify what action they will drive which gives you the indication to how often you will need it, sometimes seeing a KPI related to time on site every second week with the release cycles can help you understand if the new releases improved or not the user experience, you don’t need to see it daily as it will have nothing to the action of it.


Next, establish data governance rules. This isn't about restrictions but about setting expectations. When will the data be ready? What is the one true source of the data? Who owns it? How quickly will issues be resolved? Clear agreements create a supportive system between data users and providers. Everyone knows their roles and what to do "when the you-know-what hits the fan" (because, let's face it, it will!).


In many companies, the idea of data is that everything is important and needs to be used, but in reality, some data is less important than others 🦖. This led us to hire more engineers and analysts and increase our investment in infrastructure, but as I mentioned a few times in previous articles and my book, will more than 5 KPIs change the decision that needs to be taken? If you need more than that, restart the process of defining your KPIs; you did something wrong there!


When you start with too many KPIs it means that you need a consultant each time you need to change something in the dashboard, is it worth it? If you arrive at a meeting and one KPI shows two numbers, can you decide based on it?


During my career, I learned an important thing: it’s enough that the data consumer spotted twice an error in the dataset, you lost their trust forever, and from this point, they will check the numbers before they believe you that the numbers are right, generating a few extra hours of work for validation, building a positive trust from the beginning with a system, that can notify the moment there is an issue with the data will save you long term extra investment and hustle!

Building Scalable and Iterative Infrastructure

Building Your Data Oasis: One Step at a Time

Now that you know what data you need, let's talk infrastructure. Think fast, affordable deployment. Forget a one-size-fits-all approach. Start with a basic system that can be easily upgraded. Treat it like a product platform, evolving in cycles. Each cycle focuses on either improving infrastructure or optimizing costs.


The truth is, as a user, you don't care if your data lives in a fancy mansion (Snowflake) or a sturdy apartment building (S3). What matters is getting accurate, timely data. Your team needs to think about how they maximize the buy/build approach for the first iteration because they will learn all the time about what works well and what does not, as well as where the issues lay.


The main purpose of picking infrastructure should be to empower, first and foremost, the health metrics you picked and give open for more analysis to be done by the analyst. It needs to be something that will be easy to switch or optimize. But for me, the more important part of the infrastructure is to launch something I can test with the users and learn how it functions for them, whether they see issues or not, and as so, don’t deploy and forget until it breaks, the data people needs to go around and collect feedback, learn the use cases of people with the data and map and prioritize it in the road map.


Embrace the Journey, Share the Oasis

Data strategy is about making informed decisions for a successful journey. By focusing on clear goals, strong governance, and scalable infrastructure, you can create your data oasis – one that fuels your company's growth.


I am working nowadays on my new book, I would appreciate it if you share it in the comments or find me on Linkedin with more details about your data journey what worked for you, what didn’t, and looking back what would you have done differently?