Daniel Bailey and his analytics team were readying a data project for a key client when a choice forced itself on the group: streamline their modeling, or run out of time.
“We were under a tight deadline with a Fortune 20 company,” said Bailey, director of commercial analytics at consulting firm Elder Research Inc., based in Virginia. “We had spent a lot of time in structuring data, building features, and we had used different methods for how the data could be modeled.”
Bailey’s team included veteran data professionals and younger members who were new to the nuances of collaboration and protocol. “The senior folks said, ‘We have to cut out some approaches, because we’re out of time.’ One of the juniors – super bright, master’s degree, all the education – I could tell she didn’t know if it was appropriate for her to speak up.
“But creativity and inspiration comes from all levels of your team. Ultimately, the junior member’s approach led to finding the best model for the client,” said Bailey. “Fostering a culture of humility and integrity really helps in making sure that you are creating the best product.”
Analysis remains a human exercise. Despite the power of data software, our use of information is often shaped by our weaknesses: bureaucracy, politics, haste, simple error. These influences can lead members of a data team to varying numbers or conclusions and result in divisions within the team – a lone analyst advising one strategy and his colleagues another. The group, far from agile and incisive, bogs down in dissonance.
As big data grows more pervasive, consultants and data scientists must hone their ability to resolve such disagreements. The insights to mismatched findings and advisories – and the skills to reduce such inconsistency in the first place – lie in a mix of technology, context, and a work culture open to the viewpoints of all staff. In the age of analytics, conflict resolution has become as critical a skill for data and business intelligence pros as math acumen or deductive reasoning.
Of course, technical skills are paramount for an analytics team working with the heterogeneous systems so common to many businesses. “More often than not, you’ll end up with an IT-centric solution, an enterprise solution, tied to the corporate standard on analytics like IBM Cognos or SAP BusinessObjects,” said Don Loden of Decision First Technologies. “But the direct counterpoint is that machine doesn’t move fast enough for the business and support decisions on the fly.
“Then you’ll have a more agile technology, like Tableau or Microstrategy, interdepartmental solutions that require the analyst to interact with the data at a level that they might not have to with the enterprise data. Because of this, you can have two different people show up at the same meeting with different data.”
Loden said that conflicts of this nature often can be resolved in a reasonable amount of time.
“It’s more challenging to figure out why the (solutions) were off,” he said. “Say the enterprise solution was lagging and the line-of-business solution was more correct – is it easy to retro-fit the correct element into the enterprise standard? But deeper than that, you have to look at why the agility was not realized out of the enterprise solution.
“Having controls in place doesn’t mean things have to be bureaucratic,” he added. “But, for example, maybe there’s a regulatory constraint and the line-of-business solution didn’t know that.”
“Often, you’ll find that within a company is another data source. It just may not be where they thought it was,” said Elder Research’s Bailey. “They may not be collecting what they thought was predictive for solving that problem.”
With trust often fragile between analysts and internal partners, some organizations publish sources of information or details about analytic modeling, promoting the spirit of openness within their corporate cultures.
“I’ve been involved with large audits in telecommunication companies, and we’ve had to explain why information generated by the analysis is differing,” said Rich Chlebek of BICP. “These telecoms are in a regulated industry and, for me, it has always been a key to understand the regulations and the calculation rules of the disparate analytic systems involved.
“It often comes down to transparency,” he added. “The various staff members are more trusting of the information coming out of analysis when they know what those sources of information going into the analysis are.”
An appreciation for more nuanced skills seems to be growing in the data analytics culture, which is so often considered a world of cold empiricism and composure. “You want your data scientists to be passionate,” said Bailey, “but you don’t want the egos. It’s about creating the right culture and having humility in putting your work out there and letting them be constructively criticized so you can get better.
“You need team leaders – and teams – that have a certain level of benevolence, where they have the desire to do well for their fellow team (members) and their clients,” he added. “Finally, it’s about integrity. Even if the data doesn’t have the predictive power that it needs, you have to let the client know that.”
Joshua Whitney Allen has been writing for fifteen years. He has contributed articles on technology, human rights, politics, environmental affairs, and society to several publications throughout the United States.
Subscribe to Data Informed for the latest information and news on big data and analytics for the enterprise.