As a last lecture, let's focus on developing a data product strategy. In this lecture, we will describe main steps to iteratively defined a data product strategy that is relevant to you and your organization. Before we focus on a data product strategy, let's look at what strategy means. Although it's associated to being a military term, a dictionary search on strategy shows the meaning as a plan of action, or policy designed to achieve a major or overall aim. This definition calls out to four major parts that need to be in any strategy. Namely: aim, policy, plan, and action. Now, we are talking about a data product strategy. So, what do these four terms mean for us? When building a data strategy, we need to look at what we have, what end high level goals we want to achieve, what we need to get there, and what are the policies around the data from beginning to the end. A data strategy starts with big objectives. Notice that, I didn't say it starts with collecting data because in this activity we are really trying to identify which data is useful and why by focusing on what data to collect. Every organization or team is unique. Different businesses have different objectives. So, it's important to educate yourself as a data product builder and define what your team's goals are. To build the right product and find the right opportunity, it might be useful to find your objectives. Once you define these objectives, or generally speaking what are the opportunities for data products to build to turn data into an advantage for your business, you can look at what have, analyze the gaps, and prioritize the actions to get there. It is important to focus both on short-term and long-term objectives in this activity. These objectives should also be linked to data analytics with business objectives. To build a good data product, each company need to evaluate, have data science or big data analytics will add value to their business objectives. Once you have established that analytics can help your business, you need to create a culture to embrace it. The first and foremost ingredient for a successful data science program is organizational buy-in. A data product strategy must have commitment and sponsorship from the company's leadership. Goals for using data should be developed with all stakeholders and clearly communicated to everyone in the organization so its value is understood and appreciated by all. The next step is to build your data product team. The diversity team of scientists application developers and business owners is necessary to be effective. So is the mentality that everyone works together as partners with common goals. No one is a customer or service provider of another rather everyone works together and delivers as a team. In the short term, start with a small team who can put together ideas and minimum viable products based on experiments. Many organizations might benefit by having a small data science team whose main job is to do data experiments and test new ideas before they get deployed at full-scale. They might come up with new ideas themselves based on the analysis they perform. They take a more research level role, however, and their findings can drastically shape your business strategy almost on a daily level. The impact of such teams becomes evident over time as other parts of organization starts to see the results of their findings and analysis affecting their strategies. They become strategic partners of all verticals in your business. Once you see something works, you can start collecting more data to see similar results at organizational scale. Since data is key to any data product, cultivating a culture around is crucial to the success of a data strategy. The mindset that you want to establish is a data as an integral part of doing business not separate afterthought. Data product activities might be tied to your business objectives and you must be willing to use data-driven models in driving business decisions. Analytics and business together bring about exciting opportunities and growth. It's essential that the data across the organization is easily accessed, integrated, and not kept in silos for effective data products. So, barriers to data access must be removed and the team must be encouraged and supported from the organization's leaders in order to promote the data-sharing mindset for the company. A big aspect of defining your data product strategy is defining the policies around your data. Although it has an amazing amount of potential for your business, data-driven products should also raise some concerns and you should plan in the long-term around these concerns. Although this is a very complex issue, here are some questions you should think of addressing around policy. What are the privacy concerns? Who should have access to or control data related to this product? What's the lifetime of data? Lifetime of data sometimes is defined as volatility. How does data get curated and cleaned up? What ensures data quality in the long-term? How the different parts of your organization communicate or interoperate using data? You should probably also look for if there are any legal or regulatory standards in place for this data set. Finally, one size does not fit all. You need to adopt and aligned through iterations. Data products often need iteration on both the algorithms and the delivery interface. You have to iterate your strategy to take advantage of the experience you gain over time and need technological advances and also, make your business more dynamic in the face of change. As a summary, when building a data product strategy, it is important to integrate data collection and modeling with business objectives. Need to communicate goals, provide organizational buy-in for analytics projects, and build teams with diverse talents to establish a teamwork mindset. Removing barriers to data access and integration. Finally, iterating on these activities to respond to new business goals and technological advances will ensure you have a good data product strategy and have the right data backing up the strategy.
