Watson Says... |
 |
| None of these three use cases are more important than the others. |
 |
What do you say? |
 |
| Click here to send us your comments |
|
|
|
1. Generate Insights. In this use case, the data science team uses data and algorithms to give the business insights that ultimately improve performance.
These can be profound, business-changing realizations, like discovering a very profitable set of customers that should be treated differently, determining strategies to provide same or next day delivery to consumers, or deciding on how much capacity you need for your ecommerce fulfillment center. But they can also be smaller discoveries that still pay dividends, like analyzing data to understand lost sales, determining the impact of a recent price discount, or predicting next year’s transportation spend.
Businesses have always needed sharp analysts to answer questions by sorting through data. When you have dedicated data scientists using advanced techniques, working on larger datasets, and scraping external data, you can ask new types of questions, and even get better answers to the ones you’re already asking.
2. Create Engines. In this use case, your data science team builds an algorithm that gets embedded into an existing system -- that is, your team builds a better engine.
For example, you may have a system that helps your business with pricing. The data science team might build a price prediction algorithm, which you could then plug into your existing website or pricing system. Your users would still see the same interface that they’re used to, but now the system’s price predictions are more accurate.
3. Build Decision Products. When you need people in the business to perform the same analysis on an ongoing basis, and there isn’t a system in which you can easily embed the solution, it makes sense to have your data science team build new decision products.
For example, this could be a tool that helps determine the root cause of service failures in manufacturing or customer deliveries, or a technique that aids in understanding customer or employee churn. Any such analysis that’s repeated at a regular frequency, but lacks a natural system, is a good candidate for a new data-driven decision product.
Conclusions
None of these three use cases are more important than the others. However, if you find your data science team spending all its time in just one area, you might lose sight of other potential opportunities. This is especially true if they spend all their time on generating insights -- you may miss out on an opportunity to have an algorithm add repeated value through an engine or decision product.
Any reaction to this Expert Insight column? Send below.
Your Comments/Feedback
|
