Ditch the PowerPoint and build the data product

The death of PowerPoint has been announced countless times throughout the past decade. Yet, we are all still using it as the default office tool to share information. And why wouldn’t we? After all, it feels that the quality of presentations has generally improved in recent years.

Inspired by TED talks, most of us are now using PowerPoint to support our message, instead of reading text from slides. If used sparely, slides are a very powerful means to enhance any speech through the use of associated illustrations or quotes.

But is PowerPoint still adequate to deliver information from data? I don’t think so. PowerPoint was the right tool as long as we followed the old, linear research process: research design, data collection, analysis and reporting.

This process is however completely outdated. Its linearity isn't flexible, and the whole process lacks feedback loops. The quality of output at each stage is determined by the previous stage. There is no responsiveness built in, so that every new or amended business question requires going through some, or all, of the stages again.

What marketers and insight professionals really need today is the ability to easily consume data on demand. They need to be able to derive insights themselves to respond to ad-hoc business questions. And they need to be able to share these insights with others in the business via intuitive interfaces, without the need to create a deck of PowerPoint charts.

In other words, what marketers and insights professionals need are data products, i.e. solutions that satisfy exactly these new needs. And we, from the information and insights industry, need to provide these products to stay relevant.

So what makes a good data product then? I suggest applying the following four principles when creating a data product:

1. Easy to use. Any good data solution has to cater for non-analytical audiences. Access to the data and its analysis must be intuitive, without the need for lengthy training sessions and introduction tutorials, to maximise impact.
2.  Use intuitive visualisations. Humans are visual learners. Visual analysis helps us to detect patterns and spot correlations in the data. And we should never underestimate the importance of aesthetics: slick, modern and beautiful visualisations will certainly support adaptation and help your data product to succeed.
3.  Provide interactivity. One way to engage research-fatigued audiences with data is to make it literally tangible. Allow users to drill-down and explore the data with some simple mouse-clicks or touches.
4.  Integrate business data. Correlate the data directly to revenue or retention to make the data product immediately relevant to non-insights functions within the business.

And last but not least: keep it simple. A data product does not need to be the all-conquering, all-encompassing data hub and analytics platform, which would probably need years to develop and be outdated before it has been deployed.  

Good data products have a well-defined scope, are built in an agile manner (release as soon as possible, learn, refine) and make intelligent use of existing (often open source) technologies instead of trying to re-invent the wheel.

Frank Hedler is director advanced analytics at Simpson Carpenter