People kill a data-driven strategy, and other reasons yours is failing
Organizations might believe they have implemented enterprise analytics, but that often translates to individual groups doing their own thing or manually combining data from different sources. The reality is that few organizations have achieved the goal of connecting siloed data for an enterprisewide view that’s used to compete and drive strategy.
The gap is wide. On one end, instinct rules decision-making. On the other end, analytics is baked into corporate culture. Where an organization falls depends in good part on the people involved. We asked experts for their take on what can make or break a successful enterprise analytics strategy.
Legacy culture and skills keep data in silos
Cultural and technological barriers are often to blame, particularly in established organizations with long histories and legacy business processes and technologies.
"Getting to enterprise analytics is a multifaceted journey,” says Daniel Magestro, former research director at the International Institute for Analytics (IIA), a research and advisory firm. “Consider just one project and what it needs to go through from formation, such as ‘we need to understand our customer better’ or some aspect of that, such as how customers behave in the store, to ‘we've observed a 5 percent lift as a result of taking that action.’ The whole chain of events is very complex."
An enterprisewide analytics strategy must come from the top, with clear direction on how to get there. But management mandates often fail because they aren’t enforced or lack executive support. Magestro suggests that enterprise analytics can be most negatively impacted at the vice president level, one or two levels below the C-suite. Also to blame is a lack of visibility into what's percolating up from the lower ranks, or a limited exposure to the strategic imperative coming from the top.
As for skills, plenty has been written about the scarcity of data scientists, but HR may not understand the skills required to support a healthy analytics strategy and, as a result, make bad hires.
Enterprise analytics demands complex orchestration
Enterprise analytics requires an appropriate infrastructure and architecture to ensure data quality, governance, and security. A critical piece of that is an application orchestration layer that integrates applications to provide views of data that haven’t been seen before, such as how a particular delay in the supply chain will impact certain customers in specific regions.
"There must be a map—not a project map or a strategy roadmap, but an actual map that tells us where the servers and data sets are. We've got [to have] data maps with definitions," says Linda McConkey, managing director of consulting firm O'Keefe.
Organizations have been grappling with information silos for years, but the problem persists despite the promises of application programming interfaces (APIs) that enable interactions between software components. Similarly, analytics silos—analytics specific to an application—prevent companies from seeing the bigger picture because data sets are isolated.
Point-to-point integration connects different types of analytics to each other one at a time. This method is often used as the path of least resistance, but it doesn't inherently enable the cross-functional insights companies need to compete.
"Unless a large organization was formed recently, it probably doesn't have the infrastructure or the foresight to put data first and create a backbone that ties it all together so it flows into a central data warehouse or whatever architecture they have," says Ken Lazarus, CEO at Scout Exchange, a cloud-based recruitment platform. "Just getting the data organized and into a central place so it can be of use to anyone is a big project."
Different teams may also produce different information using different methods, which results in multiple versions of the truth.
"We've seen companies struggle with that, where a well-intended team in the supply chain function produces one view of on-time delivery and the finance department has a view of on-time delivery from a contact standpoint," says Magestro. "They have different views of what should be the same thing."
When consultants don’t provide value
Some companies choose to hire consultants, which comes with its own set of challenges. Consulting relationships may fail because internal politics hinder a project’s success, the right stakeholders aren't involved, or the project’s scope was ill-defined. Then again, a consultant may simply lack the necessary skills for the engagement.
Since organizations can't transform themselves into data-first companies overnight, they must assess where they are and have a realistic game plan for moving forward. Unrealistic expectations can contribute to short-term failures and long-term biases about what works and what doesn't.
When a project lacks a realistic scope, it can run over in terms of both time and budget. For this reason, many consultants recommend staging rollouts so that the highest priorities can be addressed first and the project remains nimble enough to pivot should business priorities change.
"Doing an assessment of where you're at today can be a good starting point for a company that wants to take a multifaceted, methodical approach to elevating analytics in their enterprise," says Magestro.
Both the company and the consultancy should be involved in an assessment because a company's assessment is often less rigorous. Conversely, consultants aren't inherently aware of everything that impacts a company's level of analytics maturity, which in part may have nothing to do with technology.
Usually, knowledge transfer takes place between the parties, but not always to the degree companies would like. Consultants may solve a problem and leave without teaching their internal counterparts what they need to know. The contract should spell out in detail the level of training required.
Also, different consultancies may be working in different parts of the organization and have different ideas about what should be done. To ensure everyone is working toward a common objective and that the objective is being met, some companies appoint an executive to oversee and orchestrate efforts.
The most successful relationships between an enterprise and its consultants are based on mutual respect. Without it, consultants have a hard time being successful and enterprise clients may discover later that it would have been wiser to listen more.
Progress is being made toward data-driven culture
The journey to enterprise analytics is a long one for some companies. "Ultimately, it's a multifaceted commitment that requires a traditional people, process, technology mindset," says Magestro. "Do we have the right talent, are we answering the right business questions, are the business leaders fully engaged in what analytics can do for them, are we doing the right work, and ultimately, are we seeing the value of analytics?"
The good news is that every industry is slowly becoming data-driven even though maturity levels of analytics’ aptitude differ. IIA assessed the maturity levels of 50 companies in 12 industries. Not surprisingly, digital-native companies are further along in their analytics strategy than brick and mortar companies in older industries, such as insurance.
“When you look at Amazon’s business processes, talent, and goals, they see opportunities that are far ahead of, say, a healthcare insurer that's had to survive a lot of acquisition activity, changing regulations, and a very complex customer base," says Magestro. "It's a completely different situation.”
In end, the entire company must understand the importance of a data-driven strategy. "People are the biggest predictor of what will be and won't be successful," says McConkey. "The wrong people will kill it."
Enterprise analytics: Lessons for leaders
- Companies must treat analytics as an integral part of the business and not an afterthought.
- Getting to enterprise analytics requires a data-driven culture where everyone is responsible for meeting objectives defined by management.
- People are the biggest predictor of what will and won't be successful. The wrong people will kill a data-driven strategy.