When we think of products iphones, smartwatches, aura rings and many physical products come to mind but rarely doesdata make the cut for a product which is a bit strange because the biggestinventions that allows for much convenience in our lives – Google search andamazon products search are both Data products.
Data is everywhere but we don’t think of itas products may even be exactly for that reason because with the prevalence ofits ubiquity, not all data is readily usable. I recently started using a habittracker with manual inputs. Just the fact that I input details of my habitsmade for little impact however when I started looking at trends and days when Iwould not meet my water drinking goals in a user-friendly format did, I starttaking my big bottle around more on weekends as my habits go out of synchduring that time.
In my view, one of the biggest successesfor a company that started off with a data product is Google with the missionto “organize the world's information and make it universally accessible anduseful. Any company that already is on a strategy to manage data in a productorganization operates with the ethos that the data products need to beaccessible, trustworthy, discoverable, and useful based on what its customerswants.
So, what really are data products?
Data product is an application, tool that servesa purpose for its customers using data as the raw material with the outputbeing a usable, trustworthy, and secure source for insights. The purpose couldbe diverse based on the business but in the physical form, a data product couldbe in following different forms.
Different types of data products 1:
- Raw data: Usually needed by ML and Data scientists or even savvy businessusers who want to further manipulate it. Data dumps or extracts in excel sheetsor uploaded in a database fall in this category
- Transformed data: used by other internal customers in the organization. Gettingcleansed data from an operational system for reporting and BI teams to use fallin this category
- Algorithm: produced by data scientists used by business for recommendationsand results required by the end user. A search engine or sentiment analysis toanswer business questions would be good examples of it
- Insights: used by business stakeholders to help with decision making.Reports, operational dashboard fall in this category
- Automated actions : Alerts and monitoring that occurs based on procedures andresulting automated actions
A data product additionally follows thesame lifecycle as a software product with key stages being identification of anopportunity, building an initial version, evaluating the impact of the minimumviable product and iterating 2. Data products are discoverable,addressable, trustworthy, self-describing, interoperable, secure 3.
Data products can also be segregated basedon the purpose it serves based on how the product organization wants to groupits data products.
Why treat data as products
Around the world, organizations have beenrapidly transforming into digital organizations however very few considerthemselves data driven 4. The way we have been setting up analyticsteams serving siloed business domains aren’t working leading to manyorganizations still feeling like they aren’t data driven.
The main benefit of using the productcentric mindset would be creating only those data products that are the need ofthe business users. Any data product creation relies on close collaborationbetween the business users and technical experts to understand what questionsbusiness seeks to answer before creating a solution for them. Secondlyiterating on the primary needs to create a hypothesis and minimum viableproduct before productizing it means it presents the analytics teams with theopportunity to learn from the first implementation to improve and put thefeedback in to next iteration thus creating a product that closely resonateswith the business need. This approach sets up for adoption of the product whichis the key when evaluating its success. Lastlythe product centric approach allows analytics to be scaled, sustained and servebusiness by forming a true foundation by meeting the business needs. Buildingonly those data products that are needed implies there are lesser dark dataproducts which are rarely used.
Coming back to the Google example theiterations to the search tool have been massive moving from just the siteresults with ads presented on side to now having results from question answerformat which wasn’t possible a few years ago. That’s the product centriciterative approach in action.
In my experience and research2building data products follows these stages:
1. Initiation/ ExtensiveDiscovery: Here analytics teams must collaborate with business users start withunderstanding the key questions business is trying to answer as well as howcurrent data is captured
2. Understand the users andtheir pain points: Any product centric approachwill be unsuccessful without interviewing the actual end users to understandnot only how they intend to use the data but what they do day to day to build aproduct that would resolve their pain points
3. Analyze/ Brainstorm/ Design: Once the user persona and their pain points are understood, theanalytics team works on the best solution with user friendly designs built in collaborationwith the business and technical experts
4. Building the mvp: Once design is completed the mvp/ beta version is released toearly adopting key users who will provide feedback
5. Iterating: Once the mvp has maderounds to early adopters sharing it with other users who are able to use theproduct for similar use cases comes next. Gathering feedback from the neweruser base and including in the product to create a more generic tool for awider audience iteratively now allows to establish a data product that can beproductized
Treating data as products is the scalableway to get interoperability and collaboration that are the corner stones oforganizational data analytics success, and this trend has only begun
References:
1. https://www.youtube.com/watch?v=-TSlAntNqf4&list=PLglmvWF46yBxlC0zbvKFOIfsJ-Pz6MxcP&index=8
2. https://hbr.org/2018/10/how-to-build-great-data-products
4. https://hbr.org/2020/05/approach-your-data-with-a-product-mindset
Photo by Carlos Muza on Unsplash