Debunking Myths About Predictive Analytics

Every company would love to look into the future to see what new opportunities could be around the corner but not all would be prepared to invest in being able to do so

Why is this when the resulting pay off could be so great? Companies can gain significant long-term benefits by applying predictive analytics to their operational and historical data. If applied correctly, embedding predictive analytics into operational data can enable companies to identify and respond to new opportunities more quickly.

For example by analysing a customer’s historical purchase patterns, companies can make predictions about the types of promotional offers and/or coupons that are likely to resonate with that customer.

The ability to proactively, rather than reactively, identify and solve potential client issues before they become widespread would also deliver significant cost avoidance to a company.Companies such as Vodafone, Match.com, Graze, Macys, Cox Communications and Meredith to name a few, are already reaping the rewards of predictive analytics in areas such as reducing customer churn rate, personalised promotions and much more.

There are many reasons why companies may object to taking the step to implement predictive analytics despite the huge potential benefits it could bring. Let’s look into some myths I have collated about predictive analytics and debunk them!

 1. Predictive analytics is easy

There are many new tools on the market that make it easier for business users to analyse large volumes of data and derive answers. The challenge is that the answers may not be worth the time it took to do the calculations. Pressing a button to run the analytics program is easy but doing it correctly is difficult. Predictive analytics requires a clear objective and understanding of the application of the analysis. For example if you are using it for marketing then a good understanding of consumer behaviour is also important. Scientific evidence is not always proof, a thorough understanding of the area for analysis is key.

2. You need big data to get a solution

The availability of a lot of information about customers and prospects can be a real competitive advantage but  if the overload of data is not processed and edited in a valuable manner, it can negatively affect the analytics and turn out to be the opposite. Businesses can reach better decisions through better data management and analytics. Even small amounts of data if correctly processed can mean a huge leverage in making investments.

3. Insights = Action

Predictive analytics, done effectively, produces insights. Turning those insights into action takes both intuition and managerial skill to gain buy-in from stakeholders and pivot the organisation. Resulting predictions must also be validated before an organisation uses them for planning.

4. Predictive Analytics is only for large enterprises

Although large enterprises count as the early adopters, small and medium sized companies are now picking up on it. For example, analysis of customer retention pattern can provide a valuable foundation for designing targeted promotional offers.

5. Analytical modelling tools require a PhD in statistics to use

Many years ago this may have been true but not today. You no longer need people with formal training in statistics or the ability to program logic. Today’s predictive analytic tools make it possible for analysts with some statistical knowledge, basic database skills, and a keen understanding of the business and its underlying data to create effective predictive models.

 6. Predictive analytics is best introduced through a bottom up or top down approach

When it comes to a new implementation, the starting points can either be that the executives promote the idea downwards or it goes upwards coming from one or more departments such as IT, Marketing, Finance etc. However a mixture of both approaches works the best.  Unless the idea is supported by all stakeholder groups, it is unlikely to be a success.

7. The time-to-value with predictive analytics is long

Traditional predictive analytics implementations have typically followed the sequence of: data scientists perform some initial research, they build hypotheses, they test those hypotheses by building predictive models. After iterative testing and re-evaluating the model, the model is put into production. Predictive analytics has evolved into a tool that knowledge workers can use every day to help them make better decisions about actions to take immediately.

8.  Employing predictive analysis costs a fortune

There are many more solutions available today both on premise and cloud based. For example, SaaS (software-as-a-service) and On-Demand software both provide quick and easy access to business analytics. Predictive analytics doesn’t have to break the bank. New software and cloud storage make it within reach of most businesses now.

9. Predictive analytics eliminates the need to know your business

The people who create predictive models need to know a great deal about the business and the processes being modelled. If not the models won’t reflect the true reality and will serve as a poor guidefor decision making.

10. Predictive models that perform well (i.e. have high accuracy) provide high business value

A model’s accuracy has no correlation to business value. A model can be highly accurate but offer little value to the business. For example, a model that accurately predicts customer birthdays holds little or no value for companies that sell to other businesses, but can be very profitable to an Internet company that sells online gift cards to consumers.

11. If you can measure it, it matters…

Predictive analytics relies on metrics many of them historical data, some from studies and so on. There’s the prevailing belief that things only matter if you can measure them. Sometimes things you can’t measure make a whole lot of difference, for example – trust.

12. You can’t start until a data warehouse is in place

One reason many organisations don’t take advantage of predictive analytics is that they believe that a heavy-duty data infrastructure must be in place to get started. Today’s tools can handle a much wider assortment of data situations. Many allow end users to get insights into their own data while working with familiar tools such as Microsoft Excel. With self-service, easy-to-use predictive analytics tools, any and all business users can leverage predictive analytics with the day-to-day data they have.

 13. Correlation = Causation

Predictions are primarily based on correlations between the data you have. However correlations don’t necessarily mean that one factor caused the other factor even though they are related.

14. Predictions are forever

More data usually makes predictions better. As time goes on, new data should be added into your predictive model to make better predictions about the future. Sometimes, the model may go haywire due to cultural shifts, demographic changes, and other events that might radically change the model.

15. You need a skilled consultant to implement predictive analytics

Predictive modelling requires an intimate understanding of what data is available or can be collected, the goals of the organisation, insights about the organisations culture, structure and market. Consultants rarely have the internal knowledge necessary to run an effective predictive analytics program on their own. You might have to invest in hiring or training employees as well to complement the implementation work of the consultant(s).

16. Analytics can tell you things about your business that you don’t already know

The results of predictive models raise awareness about things business users may already know, perhaps intuitively, but are not focused on. Predictive analytics can help connect the dots between events and behaviour, providing clarity on what or where to drive the business.

17. Predictive analytics is a black box

You pour data in and something happens in the box (computer) that yields accurate predictions. It’s an interesting idea, but not completely true. Pouring in data often generates a lot of spurious correlations that don’t really mean there’s a relationship between factors. However sometimes pouring in data and seeing what comes out is an effective first step in predictive analytics.

18. Predictions are perfect

Predictive analytics produce probabilistic estimates of the future. A crystal ball doesn’t exist to predict with complete accuracy.

​​​​​​​19. The results are perplexing

Indeed, traditional predictive analytical tools often produced indecipherable results. However today’s predictive analytical tools can demonstrate their findings in easy-to-consume visualisations that most business users can understand. For example, simple bar charts are used to demonstrate which factors are the strongest in indicating customer churn or what two products are most often bought together.

A Final Word

If you have managed to read this far, I hope your mindset on predictive analytics has made a positive shift. There are great benefits in it for you if you approach predictive analytics with a new mindset. You may also want to read an interesting research report by the Aberdeen Group on Predictive Analytics: Breaking Through Barriers to Adoption. You will need to register on their website to receive the report.

For more information about implementing predictive analytics in your organisation, contact Birchman Group.