On the other end of the maturity model spectrum is data product automation. A good example of this is the product search function on Amazon. The vast majority of products purchased on Amazon are discovered using search, and improvements to the search algorithms can dramatically enhance the customer experience and lead to revenue gains.
There are a few interesting things about Amazon Search. First, it uses artificial intelligence trained from user actions such as add-to-basket, click, and purchases—and then retrained several times per day. It also considers all the data Amazon has from years of product searches and the resulting purchases. It also considers merchandising scores, prices, ratings, and reviews, so that Amazon is only surfacing products that customers will likely enjoy. Finally, it considers shipping speed and also presents a variety of different types of items if the customer search is high level. As a high-maturity “automation” data product, it has a product owner with a dedicated team and funding, and the product itself is fully automated.
Five Tenets of a Successful Data Products Strategy
Here are five keys to building and maintaining a successful Data Products strategy:
- Working backward. When proposing a new data product, work backward from the problem or business opportunity. At Amazon, teams use a set of questions to clarify the need for a data product. Who is the customer? What is the customer’s problem or opportunity? What is the most important customer benefit to be derived? How do you know what your customers want?
- Two pizza teams. Small, agile teams—so named because they require no more than “two pizzas” to feed—can move quickly with full accountability. They should have a product owner, key stakeholders, a product pipeline, and an operating plan. They are aware of key performance metrics and have well-defined goals.
- Prioritization with business objectives. Because many analytics initiatives are not aligned with annual business goals, getting business value from the analytics is often not realized. (One global CPG customer told us that 80% of the dashboards it builds are never used). This is exacerbated by the fact that centralized data teams often don’t understand the data or the area of the business they are trying to serve. The best data product teams are highly aligned with the business and demand commitments from leadership before delivering functionality.
- Service level agreements and KPIs. The data product has well-defined service levels for up-time, data quality, and response times. KPIs such as customer acquisition cost, forecast accuracy, and basket size all roll up to meet business goals for revenue growth and margin improvement.
- Roadmap. Successful data product owners will carefully curate a backlog of features based on input from business stakeholders. They’ll conduct operational planning to request ongoing funding based on team size and value for the business.
The Data Products approach is a new way of working that raises questions about data security, enterprise architecture, and governance. The fundamental difference is that Data Products create accountability on the data product team to deliver for the business. IT and enterprise architecture teams, in concert with the CISO, then become enabling functions as opposed to gatekeepers for cloud provisioning and tooling. Data product teams have to meet enterprise requirements around data security, privacy, and exfiltration of data but have flexibility in terms of how they meet those requirements.
In summary, integrated data and analytics will continue to be a top investment priority for CPGs for many years to come. As you think about a data product within your organization, ask yourself where it falls within the maturity model, and use our key tenants as a guide to build and maintain a successful strategy.
To learn more, check out our whitepaper, ‘Making The Shift to Data Products – The missing guide for how organizations become data driven’.