A worldwide software company had lots of data which could guide focused sales targeting. The problem was bringing it all together and delivering it to the sales teams in a simple process.
JTA The Data Scientists delivered a custom dashboard to drive sales, implement new processes and algorithms and resolve the challenges.
Our client, a large Fortune 500 company sells software, hardware and services. Their Small and Medium Business division had data showing the companies about to invest in a technology. We wanted to use that data to drive sales earning a strong return on the cost of acquiring the data.
The client had run some trials using small quantities of data and the results were impressive. They needed a process to convert the data into sales therefore reaching more of their customers worldwide.
The propensity data told us when a customer was ready to buy a technology or product. Internal data helped to give year-to-date purchases and a history of sales. It made business sense to seek customers who were not only probable purchasers but were also likely to buy in volume.
We also wanted to refine the process by recommending a sales play which would be influenced by the customer’s industry and number of employees. This data was to be provided by Dun and Bradstreet but was hard to match. Experiments with probabilistic matching had previously resulted in a lot of incorrect matches.
There were many challenges to be solved at the same time and so we deployed a multiskilled team under the leadership of a JTA partner:
Our Data Engineers were tasked with building an OLAP cube to stage all the data that the dashboard would need. For this they used tabular cubes that can be optimized to keep data in memory for very fast response times.
The Data Engineering Team also solved the problem of linking together different data sources by establishing a reference data repository. This system not only stores the official descriptions for data but also manages the mapping rules that allow us to join data. Finally, our Data Engineers built all the data pathways and tasks that would allow new data to flow through the system with the minimum of user intervention. These systems also provided control reports to warn of any reconciliation problems or missing business rules.
At the same time, our Data Analysts were busy building text matching algorithms. The secret to good text matching is to validate all potential match candidates by triangulating with other data points. If we can see multiple match points, then we know the match is likely to be correct. When we didn’t have data for triangulation, we used web scraping to pull extra information from the internet.
Our Java developers were also busy writing custom code that we could embed into PowerBI reports to exactly replicate the PowerPoint designs proposed by our client. Not only did we write custom visualizations, but we also built custom user controls and export facilities so that sales teams could see their targets in Excel spreadsheets in addition to the online dashboard.