Data Product Notes | How to Build and Manage Data Product


What is data product

A data product is a capbility that meet end goal using the data.

Data product vs Data as a product

Some companies started calling data tools, analytical tools, DWH as data product. As these are capabilities that help meet a goal. However these are still software product and not data product. To avoid confusion and discuss real data-product a narrow term is defined "data-as-product". This blog and site will discuss data-as-product.

Data-as-a-product is moe precise and narrow term. It describe how domain and data owners should adopt product mindset and new data attributes will be generated to keep multiple customers in mind.

Examples of Data as product

Generating movie summary, news summary is new data. While genearting these data owner can consider multiple stakeholders e.g. movie audience, reporters, actors fan, artist etc.

Linkedin endorsement is great example of data-as-product. When a user endorse other user for skill - it help users, it strengthen linkedin network graph, it help recruiters, it help determining who are expert in field. As this attribute serve multiple scenarios and customers, it is dat-as-product.

How Data product and software product are different

Software product companies identify cpability that can serve multiple customers. They build these cpabilitis as code, package it and sell to multiple customers as product.

In data product data should be generated to keep multiple customers /use cases in mind.

How to Build and Manage Data Product

Domain and data teams should use software-product thinking. The should deterine how new data attributes/data assets will generate value in multiple use cases and serve multiple customers.

How to gain value from data product

There are many ways to generate value from data-product e.g.

  • Selling data signals e.g. movie summary, analyst summary
  • Providing signal to multiple teams in company e.g credit worthines
  • How enterprise should build in-house data products

    Learn challenges and changes - traditional software companies should handle to become data product company

    Examples of good data product

  • Linkedin Endoresement
  • Google Analytics
  • New summarization
  • Consumer price index
  • Product recommendation
  • Types of Data Product

  • Raw Data e.g. linkedin endrosement Field, tagging on Facebook
  • Derived Data e.g. Consumer price index, risk index
  • Algorithimic data creation e.g. news summary
  • Decison Support e.g. google od Ad analytics data such as click thru rate
  • Automated Decision support e.g. automated drone,