By Robin pauwels - business development manager
On this day, business leaders and managers around the world are well aware of how data has surpassed the hype cycle and how it is revolutionizing the way companies operate in every way possible. A 2020 McKinsey Global survey on the state of AI in corporate organizations, reports that 50% of the respondents have adopted AI in at least 1 of their business functions. 22% of respondents claim to have at least 5% of their 2019 EBIT earnings attributed to AI, and 48% have less than 5% EBIT earnings. These AI pioneers are taking a big lead on companies who are still struggling to capitalize on data. What separates these leaders from the rest, is viewing data as a strategic asset and follow a clear company-wide data strategy.
We strongly believe that data will affect every single business, big and small, and improve the way they work. Therefore, every company should be aware of the possibilities with Big Data and AI technologies. Luckily, this awareness is growing more and more every year. But still, many companies lack the right knowledge to effectively start adopting AI.
This is where building a solid data strategy will help you in successfully improving and disrupting your business. There are 5 key steps you need to take in order to build a good data strategy to start leveraging true value from data:
- Set clear business goals
- Ask business-critical questions
- Make strong business cases
- Create a roadmap for execution
- Execute your strategy
Set clear business goals
First of all, start with defining your overall business goals. Every successful organization should have 3 to 5 yearly goals they want to achieve. These goals should be a derivative from your company mission, vision and strategy. Goals are very important to have a clear understanding of where you want your organization to be in the next few years. They provide clarity for your people and help you in achieving your business challenges.
We suggest creating an OKR framework to set Objectives with measurable Key Results. This framework provides you a clear direction of where you want to go with your company, how you want to get there and what specific steps you need to take.
For data to be truly useful in a business sense, it must help in addressing your business needs and help you reach your strategic goals. There are many different types of data that could help you, like sensor data, images, video footage, GPS location, and much more. However, capturing all these data might not be what you really need.
In order to find the right data for your business, you must first define how you want to use data. Certain types of data may be used for some goals and other types of data may be used to achieve other goals. We see 3 areas in how data can create true value for a company:
- Data can help you in making better decisions
- Data can improve your operations
- Data can be an asset that you can sell
Your data strategy can cover all three of these areas, or maybe just one. This heavily depends on your business. Take your company-wide strategic objectives and look at how these areas might help you in achieving them.
When you have a good idea of what goals you want to achieve and in what areas you want to generate value, you need to know what data is important in order to achieve your goals. Therefore, we need to ask the right questions!
Ask business-critical questions
It is easy to drown in the vast amount of data that is available today. How much data you need to collect depends on your strategic goals or the way you want to generate value. The power of Big Data is not in the data itself, but in how you use it.
For example, when you want to sell your data for money, it could be a good idea to collect all the data you can get your hands on. If you are only looking at making smarter decisions, we suggest identifying your company’s priorities and keep your focus on the data you really need to cover these priorities.
Therefore, you need to define business-critical questions and then collect and analyze the right data which will answer those questions. A good data strategy will help you in choosing the right business questions and in prioritizing them, ensuring you use your time and resources in the most effective way.
Look at your current business areas like customers, markets, competition, finance, operations, and employees. Identify what business area you want to approve and think about the key business questions that can help you in making smarter decisions, improving operations, or how you use data as an asset. Ask questions like:
- Why are some customers not buying from us?
- How can we segment our customers?
- How satisfied are our customers?
- What are our most and least profitable products or services?
- How can we optimize our supply chain?
- What is our ecological footprint and how can we reduce it?
- What employees are at risk of leaving?
- What key skills will we need in the next two years?
Create a list with as many questions as possible where you think the answers can deliver meaningful insights to improve your business areas. When you have your list, try to prioritize them and narrow them down to your top 10 questions for every specific business area you want to focus on. Always focus on the key questions that are most important to achieving your strategic goals.
Let’s take sales forecasting as an example. Sales forecasting can be used as input on a tactical and/or strategic level. On the tactical level, sales forecasting could be used to predict sales for the next 4 weeks. This usually happens with the day or weekly based data, for a specific business unit. This forecast could then be used as input for the optimization of your production planning for the short-term future.
On the other hand, sales forecasting could also be used on a more strategic level. C-level management could make a forecast for the long term to support strategic decision making like marketing budgets, budgets to expand production capacity, provision of extra storage space, and so forth. This usually requires data on a lower granular level (monthly) as also on a global level (not per business unit, but consolidated on a corporate level).
This shows that your upfront questions are very important to know what data you want to collect and how you want to start using it. Therefore, making solid business cases is your next step in finding the right answers to your questions.
Make strong business cases
A business case is not the same as a business goal. Remember, that a business goal is a destination you set for you company. A business case can help you in reaching that destination. A good business case is much like a business plan.
Your business case will determine how you will use data and AI technologies in your organization, what resources you need for building the right data infrastructure and architecture, how you are going to collect, store and visualize data, what skills are needed for the implementation, what costs you need to take into account and how it will bring value to your organization.
Make sure that you cover following components when building your business case:
- Collection: collect your data into your platform from various sources
- Storage: store data in a future-proof and efficient manner
- Cleaning: create clean and uniform data formats
- Access: think of efficient ways for accessing data from various applications
- Processing: transform and combine your data for different business cases
- Training: train your models and see the results of different training experiments
- Hosting: an easy way to deploy and access your models through API’s
- Versioning: keep different versions of your models and allow for A/B testing
- Scaling: allow for dynamic scaling of development, hosting and training
- Hardware: scalable hardware and GPU acceleration for training models
- Monitoring: monitor the operations and performance of your platform(s)
- Testing: ensure the stability and durability of your data and models
- GDPR: adhere to regulatory privacy requirements
- Governance: ensure everybody knows where the data is coming from and what it means
- Security: ensure your data is being stored in a secure way
- Archiving: archive data that is less frequently used
- Visualization: visualize your data to make it more tangible and understandable
In this stage, it is also very important to sell your case to the people in your organization who are going to implement it or who are going to have to work with it. Many people are still skeptical about the use of AI and data. One of the biggest fears is probably the fact that they could lose their jobs because of processes being automated.
That’s why you need to instill confidence with your people that your business case will benefit them personally or that it will improve the overall way of working in the company. When your people understand the value of data, they are much more likely to incorporate it into their daily operations. This is why your data strategy should have a great focus on fostering a data culture across the organization.
Now you know what you want to achieve, how you are going to achieve it and what resources are needed, you have to create a proper plan or roadmap to start implementing your business cases.
Create a roadmap for execution
Your roadmap will be a visual representation of the specific activities that need to be executed to successfully implement your business cases. A typical data strategy roadmap is divided into 3 parallel phases:
The planning phase will cover your business case objectives and KPI’s, milestones or deliverables, timeline, roles & responsibilities, functional & technical requirements, and project management.
The development phase will focus more on tasks and processes needed for the implementation of the technical infrastructure & architecture, functionalities, APIs, data pipeline, and ML models.
The review phase will mainly cover things like GDPR compliances, security, data monitoring, quality assurance, data governance, and data audits.
All these activities will be combined into a clear plan of action which is spread over a specific timeline. For every activity, it will be clear what needs to be done, what skills and technologies are required, who will be working on it, and when the activity should be finished.
Execute your strategy
Now it is time to put your plan into action. We suggest to always execute your roadmap into specific steps:
- Build the right competencies
- Launch lean prototypes
- Measure for success
- Scale to production
Make sure you always start with building the right competencies and skills before you start implementing your business cases. You can do this by providing specific training courses for your employees to ensure they master the right technologies that are needed for your projects. If you don’t have the right internal expertise to set up these kinds of training courses, we advise you to turn to specialized consultancy companies who can provide these types of training.
When implementing your business case, always start with agile and lean prototypes to quickly find out if you can find valuable answers to your hypothesis or business questions. Starting small enables you to efficiently evaluate and pivot where needed without losing too much time or other valuable resources.
A good way to measure the success of your projects is to implement and manage OKR and KPI frameworks. These are powerful tools to monitor your progress towards your milestones and to keep track of the key metrics that define whether your business questions are answered successfully or not.
When everything goes as planned, your prototypes are achieving your targets and your people are acknowledging that the project and the data are leveraging true value, you can scale your prototype to a production-ready environment. In this phase, you are truly adopting AI in your business functions and making it a part of your daily business processes.
Congratulations! You now have successfully created a data strategy for your company and put your first business case into production. You are officially part of the 50% AI adopters and hopefully, of the 22% companies gaining at least 5% EBIT earnings.
Every business can leverage value from data, as long as they start with defining a good data strategy. Don’t start to implement AI just for the sake of AI. Always ask yourself critical questions and look for the right answers you need to know to be able to achieve your strategic goals.
If you came this far, you now possess the knowledge on how to start developing your own data strategy to leverage true value from data and AI for your organization.