Cloud Hosting

Your business has been collecting masses of data for several years, but is your enterprise using that information to drive your company forward? Data for its own sake is useless. However, when data forms the basis of a well-designed analytical process, tangible and actionable information can be revealed.

According to IBM, among survey respondents who had implemented predictive analytics, 66% say it provides very high or high business value. Predictive analytics initiatives show a median ROI of 145%, in comparison to non-predictive business intelligence initiatives median ROI of 89%.

Having a predictive analytics strategy for your company is now a commercial imperative. You have already completed the work to rationalise and connect datasets, the next critical step is to make that information work for your business.

Speaking to Silicon UK, Andy Crisp, senior vice president, Global Data Owner at Dun & Bradstreet explained how reliable predictive analytics can be used to base accurate predictions upon.

The reliability of any predictive analytics varies based on a number of factors: Firstly, how many data points you have to build your model given that the more collaborating data points available, the more effective the analytics, said Crisp. The second is how accurate the data collection mechanism for the data points is and, as a consequence, the quality of the data itself. And lastly, how much history you have for these data points and how far into the future youre trying to predict. If all these factors are considered, and the data itself is of a high enough quality, predictive analytics can be extremely reliable and hugely useful for businesses.

How reliable can predictive analytics be especially when a business is using these predictions for major strategic decisions upon? Predictive analytics needs data, but is the predictive analytics engines results only as good as the questions you ask it?

Ash Finnegan, digital transformation officer at Conga also offered this advice to any business leaders looking to improve how they use predictive analytics: It is crucial that companies first establish their digital maturity. This is where they currently stand in their digital transformation journey and how their data is currently being processed and stored. To do this, companies must first evaluate their operational model, assess its suitability, and identify any pain points along the entire data cycle. The key is to arrive at a clear understanding of how and where change needs to occur in a phased manner, to progress and improve the organisations overall operability and unify the data cycle.

Finnegan continued: True business intelligence enables organisations to take themselves to the next level. By establishing their digital maturity and recognising which areas of the operational model need to be improved, predictive analytics will be empowered. Leaders will have far greater visibility of their revenue operations and will have established true business and data intelligence they will be able to identify other areas that may need to be tweaked or fine-tuned, leaving them far more agile and adaptable for any given outcomes.

Implementing predictive data analytics is clearly how businesses can improve their bottom line, but as the Harvard Business Review Pulse Survey concludes, there are obstacles to overcome:

There are many interrelated obstacles to improving the use of analyzed data. The top three barriers cited by survey respondents were lack of employees with necessary skills or training (61%), lack of high-quality data (43%), and a company culture that tends to limit access to information (24%). While the first two challenges are significant, the third one might be the most pressing.

In fact, many of the barriers cited by respondents indicate lack of access, including lack of necessary technology (22%), an organizational structure that impedes access to or use of analysed data (20%), lack of leadership prioritization (19%), and hard-to-access analysed data (18%).

This analysis is telling as it reveals that infrastructure, leadership and even business culture can have an impact on how successful any predictive analytics program could be.

And there is an urgency to embrace these technologies. In their Technology Vision report, Accenture concluded that 92% of executives report that their organization is innovating with urgency and call to action this year. And that over half (63%) of executives state the pace of digital transformation is accelerating.

Leadership demands that enterprises prioritize technology innovation in response to a radically changing world, says Accenture. Small pilots and incremental scaling are an obsolete luxury, and the friction between research, development, and large-scale deployment must diminish or disappear.

However, the Pulse Survey concluded that less than a quarter (24%) of respondents rate their organisations effectiveness using analyzed data as less than optimum. Clearly, more work needs to be done by some enterprises to become data-driven businesses that use the insights predictive analytics can bring to them.

Dun & Bradstreets Andy Crisp also outlined clearly defined ways predictive analytics can be used: Businesses can leverage predictive analytics in many ways, but three of the most powerful areas in which to use it would be to predict risk, identify opportunities and improve efficiency.Crucially, leveraging predictive analytics on client payment behaviour can reduce the risk of bad debt thereby improving a business cash flow.

Perhaps just as importantly is using predictive analytics to analyse customer behaviour and identify opportunities to improve a products usability which, as a consequence, can increase revenue. Finally, analysing client consumption can improve business efficiency ensuring businesses have what the client wants when they want it while still reducing any waste.

Also, Ana Pinczuk is the Chief Development Officer at Anaplan, also explained: The past year has proven that predictive analytics is critical in helping companies anticipate and react to disruption. Weve seen more customers leverage predictive capabilities for everything from forecasting to territory planning as they realise how imperative that external view really is to their operations.

One area where weve seen this really take off is within the sales organisation. The pandemic threatened traditional revenue streams at a time when businesses were highly focused on cash flow and liquidity. Now with a fluctuating economy and job market, revenue leaders are dealing with high-attrition rates of their sales reps. Predictive analytics allows sales leaders to augment internal data with predictive attributes on things like profile fit and buyer intent so they can target and prioritise accounts that are more likely to want to buy from them. This makes it easier to build fair territories and set more realistic quotas so they can optimise their sales resources and ideally retain top sales talent.

Predictive analytics has a wide application across many business processes and customer touchpoints. What is clear for all enterprises is they must have a well-defined predictive analysis strategy that must be high on their agendas as we move into the post-pandemic era.

The strategic importance of data cant be overstated. When data is properly harnessed it can deliver tangible benefits right across a business. However, when data is used as the basis for prediction, new opportunities can often reveal themselves.

David Sweenor, a senior director of product marketing at Alteryx says that placing every piece of analytical data into its proper environment is critical: Automation and Machine Learning (ML) have one primary limitation: context. Without context, making insightful, timely, and accurate predictions is a challenge. While automated analysis functions are extremely effective, they are hollow without knowing how, and where, to apply these learnings most efficiently.

The current skills gap continues to be an issue for businesses looking to make the most of their data, with data prep still one of the biggest issues. On average, data workers leverage more than six data sources, 40 million rows of data and seven different outputs along their analytic journey. Much of the emphasis ML has been on the technology, not the people, and thats where failed projects are rooted.

Sweenor concluded: Another pitfall is the ethics and bias consideration. Artificial intelligence doesnt make moral judgements. It is not inherently biased, but historical data and the creators of the model could be. When using machine learning and advanced analytic methods for predictive analytics, we need to be careful that inputs dont bias the outcomes. Today its data, and not instinct, that facilitates most business decisions.

And what does the future of predictive analytics look like? For Nelson Petracek, global CTO, TIBCOthere are several strands to the development of this technology.Based on my conversations with customers and partners, predictive analytics will continue to evolve in a number of different ways: The technology will become more immersive and embedded, where predictive analytics capabilities will be blended seamlessly into the systems and applications with which we interact.

The technology will be made available to broader audiences (not just data scientists) through the use of continuously improving tools and model-driven approaches to development. Tools will broaden in capability to include not only model development, but also additional functions such as data collection, data quality, and model operationalisation into an end-to-end data fabric or data mesh (depending on your point of view).

Open data and AI/ML model ecosystems will grow, supported by technologies such as APIs, federated learning, and blockchain. And predictive analytics will drive new, emerging use cases around the next generation of digital applications, including metaverse applications (convergence of digital and physical worlds, powered by technologies such as IoT, digital twins, AI/ML, and XR) and the next generation of composable applications.

Can businesses really see the future? Its certainly now possible to make accurate guesses over a wide range of critical business questions. The data that is needed to make these predictions is embedded within every enterprise. The key is to ask the right questions.

Ana Pinczuk is the Chief Development Officer at Anaplan concluded: Business has never been more unpredictable, so companies need access to more data sources and signals than ever before to model future outcomes and react quickly to disruption.We are going to see more integrations and partnerships to access data sources and make predictive analytics and intelligence capabilities from Machine Learning and AI to advanced data science techniques more accessible to the average business user. We need to democratise access to predictive analytics, and technology partnerships are a key part of that equation.

Tim El-Sheik, CEO and co-founder of Nebuli.

Tim El-Sheikh is a biomedical scientist, an entrepreneur since 2001, and CEO and the co-founder of Nebuli, the worlds first Augmented Intelligence Studio. Since the age of 10, he has been a self-taught coder, and has a real passion for designing enhanced human experiences through intelligent algorithms. After a masters degree in Computer Science and Information Technology, Tim combined his experience in design, neuroscience, and engineering to start as an entrepreneur in online multitier system architectures in the media and advertising sectors, scientific publishing, and social enterprises.

What is the difference between business intelligence and predictive analytics?

The conventional definition of business intelligence is the application of various data mining techniques, visualisation tools, and analytics to overview a companys current position as well as its performance within the marketplace. Whilst business intelligence is about asking: Where are we now as a business? predictive analytics involves a more detailed analysis of past and current data trends to predict what will happen in the future in an educated way. However, our view at Nebuli is that modern business intelligence must also involve predictive analytics and one should not be used without the other.

How reliable can predictive analytics be, especially when a business is using these predictions for major strategic decision upon?

Any educated forecast depends entirely on past experiences and accumulated data to tell us about the future. What would predictive analytics tell us about coping with future pandemics after what we have experienced? In other words, if we state our assumptions that the future prediction depends on the previous data, then we can have a better understanding of what the analytics are able to predict.

The problem comes when people do not understand the underlying assumptions in the forecast and end up producing inaccurate predictions. That is why it is essential to combine as many data sources as possible, including those generated from business intelligence, to maximise the accuracy of any assumptions.

What are businesses most interested in using predictive analytics for? For example, pricing or product design?

Predicting the future is the holy grail of making money in all sectors! Predictive analytics are heavily used in the finance sector to understand and assess risk. The future of interest rates, for instance, is of key importance to commercial lenders, as well as borrowers and for any company that trades globally, predicting exchange rates is essential in holding funds in strong currencies.

Another example is investment markets, where predicting the price of an asset is critically important for investors. Other sectors such as sports, where bookkeepers predict possible outcomes of football matches or retail, where predicting sales of a product under a price change is critical due to market price elasticity, also make important use of those analytics.

Predictive analytics needs data, but is the predictive analytics engines results only as good as the questions you ask it?

To some extent, it is, however, getting good results from the predictive analytics engine is not that simple. It is, for example, essential to remain critical of the assumptions made from the given data and always get verification from an experienced data expert about the way these data are being used. Businesses should also have a robust quality management (QM) process around the use of data and include the outcomes in the risk register for the company.

Crucially, decisions should not be made without those elements reinforcing the validity of the data or solely based on the instincts of leaders in the business. In addition to that, including elements of behavioural analysis of the target customers and employees productivity is something that we also encourage at Nebuli to ensure we are happy with the outcome. Overall, it is more about building a comprehensive blueprint of your forecast.

What are the pitfalls to watch out for when using predictive analytics across an enterprise?

The biggest pitfall we see is enterprises believing that their data is structured and holds the answers to all of their questions about the future. Most of the time, this is not the case, and I would go as far as saying that there is no such thing as structured data! Why? Because your companys data can be seen structured if it complies with your data inputting policies even though those policies may have been around for several years.

Your predictions are then based on your data output and possibly hold newly acquired data from other channels that were not part of the original data input processes. Hence, your combined output might not match the key questions you need to answer in an ever-changing data-driven world. That is why we actively advocate enterprises to adopt a comprehensive data strategy as early as possible, which is the foundation for successful business intelligence and predictive analytics.

What do you think the future of predictive analytics looks like?

Avoid the hype! Machine learning and AI are the two key buzzwords that are being pushed around as the holy grail of modern business intelligence and predictive analytics. While both AI and machine learning algorithms can add significant value, the critical point goes back to your data. Without a clear data strategy, no matter how much AI or machine learning algorithms you apply, your analysis will not be any better. In fact, AI can amplify prediction errors and biases much further if the data structure is not scrutinised, optimised or analysed in detail as part of your data strategy.

Photo byGantas VaiiulnasfromPexels

See more here:

Seeing the Future: How to Use Predictive Analytics in Your Business - Silicon UK