Many businesses today feel data-drowned, spending 80% of their time and resources cleaning and preparing data but yet not getting the insights they need. Data preparation is seen as the biggest bottleneck to driving analytical outcomes, but it’s also the biggest opportunity to unlock new value and accelerate performance. 

Nimbus Ninety members gathered in the Andaz, Liverpool Street to discuss why they are data drowned, and to map out how to float back to the surface and speed-boat along to success.

FROM DATA-DROWNED TO INSIGHT-RICH

There are a few stats that are often cited about data:

• Less than 1% of data in any organisation is analysed.
• 90% of data that exists has been created in the last two years. 

The common themes? There is a huge amount of data, growing exponentially, and we don’t know what to do with it. We know what we want to do with it, but achieving that goal is a different story. 

“There is so much data,” emphasised Tim Lum, former Head of Data and Insights at Virgin Atlantic, our first speaker of the evening. “And we still make inaccurate insights – why?” 

He went on to list some of the key issues that hold organisations back: missing data, lack of contextual knowledge within the business, input error in the data, executive blindness to the potential of data projects, cognitive bias of those analysing it, and a lack of analytical maturity. It’s a hefty list, and many in the room were nodding along: the pitfalls Tim was describing were common pain-points for members. 

Indeed, these issues don’t just hold back data projects; they have revenue knock-ons as well. According to Gartner, poor-quality data is costing organisations (who partook in the survey) an average hit of $14.2 million annually.¹ That’s an estimated 30% cost to a business in revenue. 

PEOPLE AND PROCESS

One big issue is a lack of buy-in from the leadership of an organisation. This tends to follow a similar narrative across organisations: as one data project doesn’t deliver the right insight for the problem (due to the above challenges), executive faith is lost in extracting value from the data. 

Kate Rosenshine, Head of Azure Cloud Solution Architecture at Microsoft, drew on her personal experiences when asked why very few data science projects make it to production. “There’s no governance to allow data science or AI or machine learning to scale at the level needed; it needs to be implemented into the development cycle for progress to be made at all,” she asserted. 

Indeed, process came up repeatedly as the aspect that needed refining in order to move forward with data science projects to achieve the valuable insight that businesses need. Process is more the issue that develops into the people. Tim argued, “cool algorithms that aren’t integratable into the existing data aren’t helpful. We need people on the data team that understand the product and the business goal, in order to create data processes that are integrated into the business process."

Another big issue that was continually raised is the origins of data. Where does it come from, why have we got it here and how should we approach it, are questions that data teams should be asking themselves. Understanding the answers to these questions are vital for validating the data before it goes through preparation and analysis. Otherwise whole data projects could be skewed by ill-prepared or poor-fitting data, costing an organisation huge amounts of revenue.

Meanwhile Alasdair Anderson, Fintech Advisor & Analyst, said that “there needs to be a reason for change: everyone in the business needs to be aligned to the same goal – and that includes the data team.”

Education of the wider organisation on data literacy was advocated by many over the course of the evening. Building the foundations of literacy across the organisation enables employees from all functions access the data they otherwise wouldn’t be able to – and thus gain insight that can direct business goals. 

Similarly, it is important to have an organisation-wide understanding of the problems there are and whether the organisation has the right data to fix them. Kate argued that defining the goal of the business and then finding the right data is vital for success, but also for building trust in the existing data and insights obtained from AI.

“Finding something that is valuable and using the insight obtained is the best way to win executive sponsorship for future projects,” explained Eddie Short, Chief Data Officer at Telefonica. 

One of the biggest issues is how time-consuming every stage is. Automation can alleviate this pressure on human workers, as Eddie argued, but Alasdair warned that this must be intelligently done if it is going to save time. Taking the time to identify the stages in the data process that are ripe for automation saves time later for data teams, allowing human workers to be creative rather than stuck in processes that could be automated. 

With the amount of data growing faster than we can imagine, businesses need to rethink data strategies to get the most out of their data. Reducing the cost of inaccurate insight is vital for organisations that want to move from being data-drown to insight-rich.

REFERENCES

1. Gartner, “The State of Data Quality: Current Practices and Evolving Trends”, 11 December 2013. https://www.gartner.com/en/documents/2636315

This event was held in partnership with Trifacta, a software company that helps organisations accelerate data cleaning and preparation.