Using Data & Product to Drive & Scale an Organisation
For the most recent in our series of Engineering Leadership Roundtables we decided to do something a little different; instead of having a specific speaker, we wanted this event to be more of an open discussion about the topic.
We were delighted to be joined by Hizam Sahibudeen, SVP of Engineering at NewStore, Inc who facilitated the discussion around using data and product to drive and scale an organisation.
To hear the insightful experience and advice from the engineering leaders who participated in this roundtable – watch the full video here:
Using Product & Data to Build & Scale an Organisation
The first discussion of the roundtable focused on utilising data as part of the product problem validation stage.
Opportunity sizing is key to understanding whether this is an opportunity good enough for the product tool to build on and scale. The data needed at this stage is not always clear because it is a culmination of different data types. How you gather these data points is a critical piece in the product development stage.
When you create and launch a new product a key goal for the engineering team is to minimise as much as possible engineering efforts. A fundamental aspect to the concept of ‘reduced time to learning’ is having a tracking feature. This forms the foundation for A/B testing or dummy testing. Developing and utilising a lean manageable in-house tracking system allows you to send this data between different departments of the company to offer a more personalised experience to your customers.
The conversation then shifted towards the debate around Extract, Transform, and Load (ETL) vs. Extract, Load, and Transform (ELT). Which is more efficient; doing your transformations before you load your data or leaving your data as is and doing your transformations at a much later stage in the project. One perspective that emerged was that whenever you start changing or touching data you immediately create a new set of errors and bugs, as such if you don’t keep the original source it can be very hard to track. Alternatively, you could consider a hybrid solution. Adding one field into the data structure that is a translation of what you want whilst also keeping the raw data. Whilst this does require duplication, it allows you to have a data set for tracking as well as debugging.
We’d like to thank Hizam and all the attendees for making the event such a success. We are already in the process of planning the next one!