Speed and sophistication: Building analytics into your workflows

Start adding items to your reading lists:
Save this item to:
This item has been saved to your reading list.

Scientists at the Large Hadron Collider in the Swiss countryside had a challenge most business executives understand. They wanted to discover an invisible and elusive presence – something that didn’t even exist in a physical sense – and the only way to succeed was to watch the collisions of atomic particles travelling at a tick under the speed of light. They used an algorithm to focus on the right data and eventually showed that a theorized elementary particle called Higgs boson was there, unlocking new insight into the nature of the universe.

Business leaders are also searching for an ineffable thing – deep insight and awareness about their markets and customers that can seem invisible and elusive. Like scientists with a theory, they have massive amounts of data and need to frame problems in ways that help them gain new insight. But in business, a new insight alone isn’t enough. Companies must use data to make better decisions.

In our latest research, we started with the idea that organizations are becoming more data-driven and we asked executives how this is changing decision-making. We specifically asked about two core dimensions of decisions: speed and sophistication. Speed is how quickly an organization moves – the time it takes to answer a question, make a decision, take action, and measure the value created as a result. Sophistication is the breadth and accuracy of analytics used to inform decisions, which these days may not mean the lengthy cycles some executives have been used to, due to progress with data platforms and the ability to generate meaningful insights from them.

More than 2,100 executives gave us coordinates for where their company’s decision-making capabilities are today and where they need to be by 2020. Their responses show great ambition for change. Yet many have a tempered view on the likelihood that their organizations will get to where they need to be.

Company executives say decision-making needs to change

Where organizational decision-making capabilities are today and where executives say they ‘need to be’ by 2020

Source: PwC's Global Data and Analytics Survey, July 2016: Big Decisions TM
Q: Position the ball to a location that best describes existing and anticipated decision-making capabilities in your organization.
Global Base: 2,106 senior executives.

“Most executives see the possibilities that analytics creates, especially when combined with advanced techniques like machine learning, natural language processing and intelligent agents,” says Anand Rao, Innovation Lead in PwC’s Data and Analytics practice. “What we’re seeing here is some tempered ambition. There’s a desire to make better and faster business decisions, but often a lack of clarity on how to get from here to there.”

Questions to ask

“Data and analytics have benefitted us a lot,” says Barbara Wixom, Principal Research Scientist for MIT Sloan Center for Information Systems Research.  “But the more we see opportunities, the more unsatisfied we are with our ability to take advantage of them. We are becoming increasingly aware of all that can be done and that we can’t do it all – at least not all at once. The key  to moving forward is prioritization,”  she says. “That means thinking through what battles to fight and being strategic in our choices.”

As an executive, you might ask yourself why different combinations of speed  and sophistication in company decision-making are important. Thinking of decisions in this way can help you prioritize how to use company skills and resources and improve workflows where it matters most.

Consider what might happen if you thought about decision-making capabilities in terms of these dimensions:

If your company provides 3D navigational maps to improve traffic routes, for example, you will need both speed and sophistication. Layering and combining information from panoramic cameras, traffic conditions, and driver feedback, all in real time, is a truly sophisticated task – one that would require very accurate information, true intelligence in the moment and learning loops to help you continually refine your representation of the world.

On the other hand, if your company lives and dies by fast-moving consumer preferences, you’ll prioritize accelerated agility over lengthy data cleansing  and more advanced analytics. Mining unstructured feedback simply and quickly gives savvy staff a rapid analyze-decide-act feedback loop.

You may decide that what you really need is to cover the basics, where analysis of fairly simple data provides the right level of information for managers, with the right vision and incentives to use it effectively.

Yet, many companies face high stake complex decisions that require comprehensive and careful analysis. Think about determining how to accelerate market adoption of a new product; how fast you act is not as important as how you go about analyzing the problem. It’s more akin to strategy in 3D chess. If you hope to master the chess moves, it is important to look at numerous data sets that best represent the entire problem, so you can build more comprehensive simulations. In the case of market adoption for a new product, this may include collecting competitor pricing data, modeling consumer switching behavior, and anticipating competitor reactions.  It would also include understanding how seasonality or economic conditions of local geographies impact demand. Opportunities for companies to expand their capabilities here are great, considering the size of economic impact associated with these types of decisions.

Process and data evolve together

Finding the right approach in your own company or unit is one of the great challenges in applying data and analytics today. “The big art in this game is to reduce the huge amount of noise,” says Michael Feindt, the physicist who wrote the algorithm that detected Higgs boson and the founder of Blue Yonder, a firm that offers predictive and prescriptive capabilities to business. “What one wants is a simple answer immediately, of course, and there is always a compromise in speed and sophistication. Many companies ask what more they can do with the data they have – so they need a process that folds the data they have into their workflow, from data entry on up.”

Too often, executives feel frustration because their data collection or their people (or both) are out of sync. Processes may have gaps, people may need training, or accountability may need to be put in place where it previously didn’t exist. “You have to know your desired outcome and its associated metric from the start,” says Paul Blase, a Principal and Global and US Consulting Analytics Leader at PwC. Once you know that, he adds, you can embed data and analytics into the right places in your workflow and create the right levels of skill and accountability. 

A client in the Consumer Packaging space, for example, wanted to augment a process to improve its demand planning by using intelligence from its representatives in the field. The hard part was establishing a practice where, instead of the reps working on their hardcopy charts, they digitized their workflow. “Once an algorithm is in a rep’s handheld device, you can increase the sophistication of your data collection and analysis all day long – and you can build upon the model, changing the algorithm to improve the accuracy of demand planning.”

Feindt stresses the importance of process and the importance of continually adapting to change. “Once it is clear to people throughout an organization that there is value in data, the quality of the data becomes better quite fast,” he says. Process and data evolve together. “The environment changes as the data changes – as it grows and improves – and so should the algorithms. Being data-driven means you won’t use the same algorithm you’re using now in two years.”

“To me there’s a tremendous advantage and value in getting the equation right,” says Blase, who talks of building a data-driven organization by structuring decision-making around process muscle. “It takes a long time because you’re dealing with people, culture, organization and change, but the companies that are starting now, when some of the real promise of AI and machine learning and automation is taking hold, will have a much easier time getting it into workflows where it can matter in ways we haven’t anticipated yet.”

Four models for data-driven decisions

You can model your program after four basic needs.


Thinking of decisions in this way can help you prioritize how to use company skills and resources and improve workflows where it matters most.


Contact us

Anand Rao

Global & US Artificial Intelligence and US Data & Analytics Leader, PwC US

Follow us