In recent years, I have worked with organizations in many different industries—banking, pensions, higher education, real estate, research, and pharma—as well as clients from different countries in Europe and North America. Out of these companies, only a few have succeeded in a true data-driven transformation. Most struggle just to get incremental changes through.

In my experience, I’ve seen two paths taken towards a successful data-driven transformation: the evolutionary path and the radical path. I’ll explain what each entails below.

The Evolutionary Path

Many organizations are working with an existing business model that they need to maintain. For these companies, they may be applying data as a way to optimize this business model. This path is a safe one to go down, but change may not actually happen. If you’re in this situation, it’s crucial that you follow these five steps:

AdobeStock_204106702

1. Define where applying data will benefit the most and go for it.

Apply a data strategy tool and find out where your most urgent business needs can be met with the lowest degree of complexity. A good place to deploy the first data science project is often in the marketing or customer oriented functions in your company. The benefits in these departments are quick and obvious, so you can gain momentum for the rest of the organization. 

2. Make sure you invest in enough data science capability.

It is crucial that you gain a critical mass quickly by either hiring enough data scientists or engaging with external suppliers. The most important first hire are data science business translators who can maximize impact straight away. It can be difficult to identify the specific technical skillset to hire before you have executed first projects or experienced commercial success. Eventually, you’ll need to hire the right technical staff and build rapidly while you have momentum.  

3. Conduct small experiments in your organization where potential benefits are greatest. 

Keep the emphasis on fast learning, success, and failure. Don’t invest too heavily on long-term projects, systems, and backend before you have a proven impact. Use an agile approach and make sure to adapt along the way. 

4. Develop a capture team that enables you to implement the results from experiments as soon as possible. 

It is a waste of opportunities and delegitimizing to your data science strategy if you don’t implement. It is crucial that your data science strategy gets the credit in your organization—and to do so you have to show results. Enable sufficient resources to deploy the good results of the experiments in the business lines. Experiments are great, but real impact is even greater.

5. Measure the impact on sales.

Apply control groups where the data-driven approach outperforms the traditional approach. I have seen good examples of this in banking and telecom. If you are able to show on a weekly basis that the data-driven approach outperforms the traditional approach by 186%, the organization will soon be convinced.

The risk of taking the evolutionary path is that you end up not moving fast enough. Your traditional business model will prevail if you don’t inject rapidly. 

The Radical Path 

AdobeStock_187224591

The radical path is challenging, but when successfully executed, it is also the most rewarding. This is the path that can get you to potentially disrupt your industry. However, you also risk disrupting yourself. If you are not a startup, you probably need to earn money while you identify your new future. Therefore, you need a strategy to keep your existing business successful while reinventing your business model from scratch. If this is where you’re at, you should take the following steps:

1. Build an organization independent from the mother company. 

In order to develop a radically new business model while maintaining the old one, you should consider setting up an independent organization that is able to do all the things that you don’t do in the mother company. An example of this is Leo Innovation Lab, an independent unit established by Leo Pharma. They act completely on their own, don’t depend on legacy systems, and are in business to disrupt the pharma industry. They develop completely new data-driven business models and also digital products.

2. Focus on how data delivers impact and value before you focus on backend infrastructure and models. 

Great ideas should always be tested in the market. I have met several tech startups who forget the importance of sales. They receive initial funding and have a great idea, but they run out of money. It’s a delicate balance. On the one hand, you need to identify interest from clients and consumers before you develop the technical solution, and on the other hand you need to be able to deliver as soon as the clients are there. Get internal or external resources to enable you to understand how to develop your business model in the best possible way.

3. Avoid being locked in by systems, programs, and licenses. 

If you can, aim for open-source software and take advantage of the cloud. One of the benefits of living like a startup is that you are not bound by the legacy IT infrastructure. This makes you flexible and agile. You can save your time and resources and instead deploy simple solutions and have them grow in smaller puzzles. When you are not relying on a system to fix everything, you can easily change tools and policies down the line without going through bureaucratic processes. You also don’t get stuck with having to wait six weeks to get a new template in an inflexible system handled by a central IT department, when you can have built it instantly in a much cheaper and better performing Software-as-a-Service solution.

4. Make sure that you cover the entire data science skills value chain.

Your people value chain should be comprehensive to maximize your chances of commercial success. The value chain contains four main categories: 

  • Back-end developer/IT infrastructure skills: This skillset includes identifying, extracting, and processing the data needed to solve your business issues.
  • Programming skills: This skillset includes building quantitative models such as machine learning models, and development so models can be deployed in production.
  • Front-end developer skills: This skillset includes creating visualizations and dashboards to make the analytics easily applicable to the people in the organization.
  • Business data transformation skills: This skillset includes developing data strategies, identifying value propositions, change management, and business development around data science–a crucial part to making sure data generates financial value to your organization.

5. Keep an eye on your business KPIs.

Don’t waste time on developing a business plan. Instead, think, execute, test, revise, and capture benefits early on in every development project. Identify and focus on the most important KPI, and get rid of any that don’t have traction or generate value. In some companies it might be turnover, customers, market penetration, and funding. In others, it might be users, paying users, and conversion rates. Whatever it is, choose the most important things for your own business and make sure to communicate the progress to your main stakeholders. 

Conclusion

In order to determine the most opportune path for your business, you should ask yourself some key questions. Will your existing business model survive in the long run through a redevelopment? If yes, you can go down the evolutionary path. However, if you risk getting fundamentally disrupted, you should go for the radical path while keeping the mothership on track. The other question relates to your clients. Will you risk disrupting your clients’ market by developing new products and business models with data? If so, you will not be able to keep them on board and continue down the evolutionary path. Instead, you will end up staying in the status quo, in which case you should also aim for the radical path.

 

Kristian Mørk Puggaard
Author
Kristian Mørk Puggaard

Kristian Mørk Puggaard is Managing Partner in Copenhagen and Stockholm-based DAMVAD Analytics where he advises organizations on developing and executing data science strategies with a business impact. Kristian has worked for large MNCs such as The Boeing Company, Novo Nordisk, ATP Group and Skanska as well as for governments, investors, and leading universities across Europe. Kristian holds a M.Sc. in Political Science from Aarhus University and Institut d’Études Politiques de Paris and an Executive Certificate in Global Management from INSEAD.