Data Culture is Vital For Organizational Growth

Corporations are digitizing just about anything and everything they can. They are realizing that data is the lifeblood they need transparent and real time access to with a level of analytics that is going to generate almost all of the value. For this to happen there has to be a data culture that reigns supreme above all.

I have always believed that the best sterilizer on the planet is sunlight. Great ideas surface when everybody sees what everybody else is doing and talks about it. That’s the essence of innovation where value gets created. This can only be enabled by data. So, in essence data is sunlight. 

An important aspect of moving to an analytics or data centric organization is the health of its culture. As an analytics leader I have realized that data culture is difficult to bring about - it’s not like a browser bookmark that can be imported from one organization to another nor can one dare to impose it. Most importantly a data culture should also not be isolated to experts or innovation groups. Organizations should strive to achieve a culture of employee engagement, purpose and connectivity so that the data can effectively support them. So, what can I share about data culture that can help you develop your own organizational analytics strategy?

The first thing to remember is that data cannot be isolated to a cool tech project. It’s not an edge experiment your division’s analytics team does in a vacuum. Remember why you are collecting and analyzing all this data? Not for being cool or for showing that you are science friendly but instead for making better decisions. A healthy data culture implies better decisions. You build a data culture because it’s good for your business and you simply cannot survive in the long term without it.

And because of this reason, it’s vital that executives be honest about the big business problems they are facing and work on solving them. It’s not enough to just collect data from around the organization and build up a massive data lake. Quantity means nothing when you don’t know what you are trying to solve? I have seen companies get carried away with creating industrial scale data lakes. There is so much data pouring in nowadays that soon your wish will be fulfilled and you will find yourself the proud owner of a data ocean. My recommendation is to avoid starting at the data and analyzing it to determine the business problem it might be solving. Instead figure the problem out first, then go look at the data.

What do you want to achieve? What results? Start there then look at your data to figure out what type of analytics you might need to create the types of insights that might reap the desired results. Then immediately report these results to your team or customers because further feedback and ideas from them will only make your model more robust and help improve the model’s effectiveness.

I tend to start in departments where decisions are already being made. I review who makes them, the processes and systems used, time and effort taken, how the reports are produced, what data is utilized etc. I also try to determine any gaps in the data and how to plug those gaps to get the type of insight or decision I am after. Sometimes it’s as simple as removing bottlenecks in the current process. 

It’s also vital that data and analytics driven culture start all the way from the top i.e. is mandated by the CEO and Board. This does not mean the CEO makes the occasional speech or a passing remark but instead a continuous discussion between executive management, the Board, CEO and the division heads responsible for the various analytics initiatives across the organization. 

It’s important to be aware that in the multi-billion-dollar enterprises, especially the Fortune 500, CEOs are mainly concerned with the largest data initiatives that will bring about billion-dollar changes. So, try to refrain yourself from presenting your smaller day to day projects that may be equally important but not worthy of a discussion with the CEO no matter your urge to engage. In our case we keep it simple - here is the money we got, this is what we are doing with it, challenges if any, how we are spending it and what the expected return is going to be over a period of time.

Praises from senior management for intelligent analytics driven decision making goes a long way in encouraging staff to think more analytically. This is a big contributor towards building a data driven culture.

Board education about data and analytics also goes a long way. Let your board ask as many questions as they can about data, its potential, analytics, ML, AI etc. This helps especially if you can show examples of cases where real value has been created. The dialogue is an essential part of building a data culture.

Another aspect of creating a data centric culture is to ensure that it is not top-down. The demand for data and analytics-based decision making should always come from the ground up. Adoption from the grass roots is essential to create competitive advantage. In our organization we have built a data culture that essentially acts like food where you feel you are missing something if it's not present. Every team meeting, discussion and conversation involves data as a key factor. If you do not do this and focus instead on building a data product that’s entirely top down and without input from all levels of the organization, you are headed for disappointment.

You also need to get people to believe in data and the value that can potentially come from its effective use. Creating this culture is perhaps the toughest thing to do. To change behavior, one needs to have a common platform accessible by everybody. With time people start believing in and internalizing the analytics platform. When this happens, it’s a game changer.

The controls around data are also important in a data-cultured organization. Policies around what is and is not allowed. Often organizations without realizing the value of their data tend to freely share it with outside vendors or other organizations for discussion or analysis. This is a mistake. Data needs to be stored in a repository that is encrypted and where others can be “invited in” to analyze and experiment. In fact it should be setup in a way that the data becomes useless the moment it leaves the organization’s repository much like data that’s stored on your smartphone. Data needs to be treated and protected like it is gold. 

Also remember that when you start to analyze data and generate reports via tools like Qlikview and PowerBI or create more advanced models using machine learning, the processed data needs to be properly supported. That means there is a cost for that support in the form of retraining, new hires, capability building, software licenses, day to day monitoring etc. You need to make sure your data stays in top form. A slight glitch and you could lose big money. You can’t build a data and analytics platform and then expect staff to start using it instantly. Your staff will need to be retrained at the very least to use such systems. You might also need to supplement with new hires or part time assistance.

I have seen several companies try their hands at building a data ecosystem but without a robust foundation to support it. They fail. Companies that create a robust foundation for data infrastructure, but do not know how to be creative with their data usage also end up stagnating. So, the key is to do both - build a foundation as well as be creative in how you build your models and use the data to make decisions. Otherwise you will not grow.

The other important aspect of creating a data culture is people that can balance both operations with data science, bridging the gap. If you create amazing algorithms that can recommend tactics and strategies, but the managers on the ground cannot understand or implement them, your data becomes useless. So, key here is to remember that your change agents for success can never be digital natives but rather your staff on the floor. It is imperative that you involve the floor staff in your data projects each step of the way.

Think of it this way. If you are developing annual forecasts for a line of business you have to assume the people that run the business are intimately familiar with and probably buried in it. So how do we bring about data driven change in such a situation? We find a researcher or someone inside of this organization to be our change agent, our point person that is open and willing. This is the person that will interface with the business people and translate between the two groups. S(he) becomes the interpreter. In this equation if we build a machine learning algorithm with several features and release them as a tool, our “guy” will need to adopt and internalize it first. Only then can s(h)e train the rest of the business crew on how to use it. With time there will be ownership that evolves allowing them to internalize the project as their own thus adopting it into their day to day business.

Let’s talk about security and privacy of your data. It matters. If you give your data to a vendor that runs some machine learning on it, you can’t be sure those algorithms and your data are not shared in some form with 30 other companies? In fact, I would argue that is exactly the way smaller analytics companies get more business from larger corporations - by showing these corporations how successful they’ve been in building new models [on your data]. So it’s vital you protect your crown jewels and store all your data inside your own platform, allowing outside vendors access to the platform for analytics and modeling purposes but then rendering the model and data toothless the moment it leaves your platform, much like your data becomes useless if it's stolen from your iPhone. You need to have control of your data in-house and build the analytics capabilities in-house. Start with consultants if you have to, but eventually bring it all in-house.

A data cultured organization also thinks about how to word the vendors contracts. I inherited a data contract once where a vendor was not obliged to return back our source data nor the enriched processed data! Further, the vendor could also use our data in their larger data lakes to support their other products and projects. We want to make sure we do not fall into such traps and paper the data agreements entirely in our favor. Your data is yours, nobody else’s. The enriched data is also yours, nobody else’s. If you find yourself compromising on these two fronts, then you should rethink working with that vendor since you are giving away your competitive advantage entirely. As a Chief Data Officer, I make sure that every single project that involves the transition of data from our internal servers to an external environment or even the Cloud goes through my office and we have protocols in place to ensure review.

Lastly, make sure you hire the right balance of people. It’s all about the talent, both in-house and new hires. I don’t hire PhDs just for the sake of it if I can find good employees within the organization that can work just as well with numbers and math. I look for capabilities and not necessarily industry expertise. If I find an awesome software engineer from outside my industry, I prefer that to an average one in my own industry because I know the amazing software engineer can create immense value for my company. It’s all about the skills. Remember if you only recruit from within your own industry, you should never expect to be better than anybody else in your industry! I look for skills and people that are quick to learn a new way of operating, can manage products effectively, are able to interact with colleagues, customers and clients and are smart thinkers. These are the people I have found to be the most effective in growing a company.

I prefer teams where people are smart and have high integrity at a minimum but then are diverse in thought and experience, strong with technology, understand business processes, maybe even have some subject matter expertise, who can communicate effectively both in writing as well as verbally and are able to create a healthy tension in the team. 

In sum, it’s important that an organization’s data culture is attached at its hip to its strategy and operations so that analytics projects meet expectations. From my perspective whenever I have seen an organization come alive and grow leaps and bounds it has usually been when its people have embraced data and analytics at its core. Add data culture into this equation and you are all set to ride into the future.

Other Articles by the Author on Data, AI and Analytics

AI at Scale - The Key to Organizational Growth

Building Effective Data Science Teams for Advanced Decision Making

How Can AI Help the Airlines Depart and Arrive on Time?

Other Articles on AI, Data and ML