Predictive data: valuable insights to guide business strategy

22/08/22

Using algorithms, data analysis and visualisations

Every company analyses and looks back. The annual accounts tell us what went well and what did not, and how results were achieved. "But looking ahead and establishing relationships between different data streams is something that many companies are still failing to do," says Miguel Brahim, a data analytics expert at PwC. "They often use Excel to set budgets and make cash flow forecasts. The use of algorithms, data analysis and visualisations is often not yet widely implemented, despite the fact that these instruments provide valuable insights to guide business strategy."

Miguel Brahim - Data expert at data analytics PwC

Faster decision-making on data

The pandemic, combined with stock market trends and technological innovations, have meant that the market changes quickly and frequently. This creates uncertainty in many companies. "New business models are disrupting industries completely," says Brahim. "Look at Uber and Mollies. Their rapid growth requires faster decision-making on data so as to generate new insights. These companies are very good at identifying risks and seizing opportunities because they anticipate figures and look ahead to what is happening in the market. Because of this foresight, a CEO and a CFO will know exactly what to focus on. Take the retail sector during the pandemic. Some companies already knew what problems would arise in the supply chain. They acted quickly and switched from physical shops to online and 'click and collect'."

Finance responsible for data and predictions

Predictive analytics essentially means gaining the insights needed to make future predictions based on an analysis of historical data. Finance will have a major part to play in this. "Because finance has all the data, it is best placed to take responsibility for that data and the predictive side of it," says Brahim. "This way, finance can help other departments. For example, the purchasing department, which uses predictive analytics to discover certain patterns and therefore to find out which products are needed in the supply chain and when. Predictive analytics is in fact the last piece in the jigsaw of existing fixed systems that employees are already used to working with. But now they can see what they are doing in real time."

Reluctance to adopt predictive analytics

Software for predictive analytics is not a service that companies just go out and buy. Brahim blames the reluctance to adopt predictive analytics on three shortcomings: "To make good analyses, you first of all need good data. In many companies, it is simply not available or its quality leaves much to be desired. Secondly, you need the right people. A data scientist models data but doesn't know which insights are needed for the business. For that, you need another professional. Finally, knowing how to use technology is important. Companies often don't know what technology they need or how to use it to produce good analyses."

"When we go into a company, we don't know what we'll find in terms of data," Brahim continues. "Is there any data at all and, if so, is it actually usable? A company must be willing to invest in data to see what is needed. Predictive analytics is therefore a bespoke service. It begins with bringing the right people together and having good communication. In transition processes, you often see that employees are afraid their job will change. But that change, for example the use of artificial intelligence, will actually make their job more interesting in terms of content. But you have to communicate that clearly."

Exponential progress through sharing use cases

An important factor in ensuring the success of predictive analytics is the wider sharing of use cases. Brahim notes that these often remain centralised. "It is precisely the reference points - the comparable characteristics in use cases - that are also very useful to other departments. Adopting successes and sharing them more widely with different departments really makes for exponential progress in the process," says Brahim. For companies that do not opt for the entire process but still want to have insight into data, a "self-service analytics" platform is an accessible solution in which companies themselves perform analyses based on specific data. "We can then help them unlock their own data, make the logical connections and get insights that are relevant to business strategy."

Predictive analytics community

It is important not only to be willing to share data, but also to share knowledge about data analysis. Brahim believes in creating a predictive analytics community. "A community where different disciplines come together to talk about data analytics as a solution. Each discipline will be able to extract a different useful insight from possibly the same dataset, resulting in different stories about data with a different purpose. With more and more work being cloud-based, the datasets are accessible to everyone and everyone can act on them in real time."

However, Brahim does have reservations about this accessibility. "There's some friction between cybersecurity and data quality. After all, you want quick access to sometimes a lot of data, but it has to be secure. Our CEO Survey also indicates this urgency." Another trend is ethical AI. "Algorithms cause a lot of discussion when it comes to discrimination", concludes Brahim. "How do you prevent bias creeping into that data? And because data analysis is often used to analyse risk and detect fraud, you have to be careful not to target certain people unfairly. It's therefore essential to build trust within the context of algorithms, data and data analysis. Consequently, it is a good idea for companies to set up a special department that deals with the correct use of algorithms and data."