Successful Data Strategy Roadmap
A digital strategy is the starting point of every digital transformation. Here, the use of data is an essential component. If your data is in order, it will help you to improve your customer approach and your business processes. It not only guides the development and use of new technology, but sometimes even the use of new business models. But how do you ensure that your data actually contributes to innovation, growth and more efficient operations?
This article is the second in a series of three in which you can read all you need to know about activating data. We not only look at the strategic and technical aspects of data activation, but also at the mindset and culture required to extract as much value as possible from data. In this article, we tell you what a roadmap to a successful data strategy looks like. As we mentioned in the first part of this series, data is only valuable if you activate it. You have to get your data to work for you and you need to ensure that it delivers added value for your organization. How do you identify a use case that will allow you to take the first step and how do you demonstrate that value is actually created?
The roadmap that will help you do this is based on an approach that we call “Grow Live”. In this approach we apply a number of core values and principles. It all starts with smart, innovative people. Experts in the field of digital transformation and data activation (data engineers and data scientists) who, together with business experts, achieve meaningful results. The business determines what value is and sharing knowledge is essential. Given that you can’t get from the basement to the attic in one go, you need to start small to end up big. Then you repeat the steps. During each step, you pay attention to business, data and information management, and governance processes (including privacy and security). These are aspects that you can’t just add on later.
Four focus areas
During this process, there are four areas you should focus on: business, data, architecture and validation. A checklist will help you with this.
- Business involves aspects like what the challenge is and why you need to resolve it, what the desired result is, and who the stakeholders are, what is in their interests as well as what the related user stories are.
- Data is all about what data you need, where it is stored, its quality, and whether it is actually available and if it is accessible.
- Architecture involves examining the platforms, systems, interfaces, connections and components that are used.
- Validation is what you ultimately use to test the feasibility of the combination of all of the above. The result? Demonstrably valuable initiatives that you can plot against each other on a graph based on their value and complexity. This allows you to start with the initiatives that in relative terms are the most valuable and the least complex.
Value and complexity
The roadmap starts with inspiration sessions, with searching for and finding a business challenge that you can solve with data. Using customer journeys and use cases, valuable initiatives are then identified. These initiatives are compared on the basis of their value and complexity, which means you can then prioritize them. The use case with the highest priority is the one you tackle first. Next, you perform a quick scan to see if this use case is feasible. The quick scan includes a proof of concept. At the end of the quick scan there is a go/no-go moment. For the feasible use cases, you then demonstrate whether or not they will deliver value. This is followed by another go/no-go moment. Development is then started for the valuable use cases.
Roadmap in practice
We introduced such a roadmap at a Dutch brewery. The organization had a large amount of complex data that was fragmented across different sources, locations and systems. The brewery wanted to cut procurement costs and increase efficiency in the supply chain. We started with a number of business workshops in which, together with the various stakeholders, we mapped out the issues. How do the procurement processes run, what are the challenges and where are there opportunities for improvement? For example, are you buying too often? Are you buying too much? Who are you buying from? Could you have done things differently? Together, we also set down several questions related to the supply chain. What components do you need, and when and where do you need them? Do you know what you have in stock and where it is? And can you find components when you need them?
This led to a stakeholder analysis with user stories, as well as a use case matrix and a roadmap. The proof of concept showed that we could generate an overview and insights using big data techniques and smart algorithms. The proof of value showed that this would deliver value for both procurement and the supply chain. We then got started with a basic platform. After that, we could work iteratively and incrementally on realizing benefits and capabilities based on the identified use cases.
You will inevitably run into challenges in this type of process, especially at the beginning.
- It’s often unclear which problem you can best address. Complexity and feasibility are key elements. Start with something that can actually be done and that also delivers value.
- Alignment between the business and the data team. This alignment must not be lacking and must continue throughout the project.
- Data (availability and quality). At the start of a project of this kind, data is often not sufficiently organized, not available everywhere and not of a high enough quality.
- A lot of technical fuss and bother. For example, is a data processing agreement necessary and if so, has it been signed? It might also be that you can’t access the data because something is in an industry – or other – system.
- Expectations that are too high. The absence of a fail fast and learn faster mindset can kill a project.
- Translating experiments into products. Production is a very different environment than the ‘laboratory’ environment you’re used to working in. There is less control and there are factors that could be disruptive. For example, changing IT that causes your data feed to stop, or a changing world that forces you to retrain your model.
And then you have the data-specific challenges.
- Fragmentation. There is data everywhere, the data formats differ, the data quality and extraction method are both uncertain, and the meaning does not match the extraction method.
- Data collection and data quality. You need to know what your sources are and what controls are in place.
- Data security, compliance and data anonymization. You need to address these processes when setting up your environment.
- Data sharing. Again, to avoid problems, this needs to be addressed when the infrastructure is built to ensure you can execute a project of this kind.
- Storage capacity and speed. The lack of either of these is a problem and if both are lacking it can be a real challenge.
It helps if you look at the project as a series of small steps. It’s also always a good thing to recognize the project’s urgency. If you don’t do it, someone else will. If you do, you’ll stay ahead of the curve. Do you have a data strategy with a business problem and the necessary data? Then it’s time to work with your organization to figure out some use cases. Prioritize them based on their complexity and value to your organization. The use case that seems most feasible and adds a lot of value is the right candidate to start with. You perform a quick scan of your organization, your data, your architecture and your connections to find out what is working well. Using the proof of value, you see if you can find connections and what value they deliver.
In the next article, we move on to operationalization and look at how to bring the concept into production and how to keep it in production.