The Business Value Potential in Self Service Data Blending

David Triggs

When we think about creating value with big data and analytics the focus is mostly around the work of data scientists and analytics professionals. Of course a variety of business and domain expertise is also needed, and the business and domain experts frequently are just as interested in better understanding the data as the data scientists and analytics professionals, and they are often the ones with the most perceptive insights to offer. This can lead to frequent requests for reports, dashboards, and visualizations that can burden even the most capable analytics professional, and worse potentially result in their being perceived as a roadblock to creating business value. As a result, especially in medium and larger organizations, it is now more important than ever to provide effective self-serve access to data visualisation and basic analytics for business and domain experts who may not have the time or technical expertise to extract the data from different systems themselves.

Providing self service access to reports and visualization capability to LOB (Line of Business) has been one of the factors in the success of software companies like Tableau and Qlik, but based on survey results quoted by Alteryx CEO Dean Stoecker, 40% of organisations have 5-10 data sources for any analytic workflow while only 6% have data all in the one place. So mostly data blending is needed before Tableau or Qlik can provide self-service reporting or visualization to business analysts and other business domain experts.

Stoecker argued this was a factor in the continued extensive use of Excel for data blending and analysis among business analysts and others outside the IT organization. He argues there are two problems with this, first Excel is brittle when used in this way as it is easy for errors to creep into the data manipulation, but more importantly it is time consuming, resulting in significant opportunity cost.

How significant? Stoecker points to the experience at Home Depot. With $76 billion revenue, 300,000 employees, and 2,200 stores, Assortment Planning required analysis of around 169,000 distinct items or SKUs. The huge number of permutations made this a difficult problem and even after Home Depot had spent millions of dollars on custom systems they could only effectively analyse a small fraction of these SKUs. Over one weekend Charles Coleman, an analyst in Assortment Planning at Home Depot, downloaded a 14 day trial copy of Alteryx and started to solve the problem himself. In around 30 days he had ingested all the required data, and went on to achieve a 2-4% increase in top line sales and $177 million in gross margin improvement; in addition to being named employee of the year. Stoecker says this rapid time to value is not unusual, and claimed 92% of customers saw time to value in a few days.

At Home Depot Coleman may well have had the skills to access most of the data, given sufficient time and access. In fact survey data mentioned by Stoecker indicate lacking skills is only a barrier for 18% of respondents, 52% sited too many tools, 37% have to wait for other people, and 30% can't access the required data. This is a significant issue for not only business analysts and other business domain experts, but also for data scientists and analytics professionals. One of the secrets of getting value from data is that its generally not having the best algorithm but rather the best data; and this is often also the most time consuming part of the process. Most enterprises have huge and growing amounts of data - the challenge is finding the best data for an often long and growing list of questions for which the business values answers. Success requires being able to quickly blend a range of data, firstly by business and domain experts to gain the initial insights from the data, without being dependent on others for data integration, and then secondly by data scientists and analytics professionals to blend the required data in the limited time they have available for each analysis.

Even before being directly represented in Australia Alteryx has been successful in the Australian market, with over 50 Alteryx customers in Australia; including Bunnings, a business similar to Home Depot, who have used Alteryx for the past 3 years. Alteryx also fits well with the trend, seen globally including in Australia, to shift workloads to the cloud. For example Amaysim, a Mobile Virtual Network Operator (MVNO) with 700k customers across Australia, but only 3 people in their BI team, deliver the analytics their business needs through an AWS hosted solution based on Redshift with Alteryx and Tableau. Amaysim BI Manager Adrian Loong says the solution has delivered speed in solution implementation, for example implementing a financial audit workflow in 2-3 days when the team were still new to Alteryx. The solution has also helped deliver self service access to information with all their management team on Tableau. In addition it has also assisted the BI team by helping collaboration between different Amaysim teams and building trust in analytics results through better visibility into the analytics analysis, including helping in collaboratively building workflows.

This is why, despite Alteryx being perhaps best known for expertise in spatial analytics, and also having capabilities for predictive analytics, that CEO Stoecker is focused on the opportunity to apply Alterix’s drag-and-drop interface to data blending to support the capability of self-service service data exploration and visualization tools like Tableau, especially for business analysts and other LOB domain experts; but also to speed exploration by data scientists and analytics professionals.

David Triggs, CTO, BigInsights