Predictive analytics moves beyond traditional BI tools to advanced statistical models and machine learning, giving insights that traditional tools are unlikely to discover. According to Gartner, by 2018, more than half of large organisations globally will compete using advanced analytics and proprietary algorithms, causing the disruption of entire industries. With 2017 now upon us, it’s time to start thinking seriously about predictive analytics or risk getting left behind.
A starting step is to think strategically about what you want to predict in line with your business objectives. Depending on your industry and business, there are a number of powerful ways predictive analytics can help you market smarter. Below are five practical uses:
What would happen if you did nothing? What would happen if you didn’t advertise on TV? What would happen if you didn’t advertise at all? Measuring the effect of doing nothing is always an option when it comes to predictive analytics and is often forgotten.
Measuring the effect of doing nothing allows you to create a baseline to measure against. It gives you a benchmark to genuinely measure uplift and decline as a result of your marketing strategy. For example, if you wanted to stop advertising on a particular channel, you could predict whether that would result in a decline in sales.
A big mistake a lot of people make in predictive analytics, is in measuring the success of campaigns. For example, if you spent 50,000 on marketing and made 150,000, you might fall into the trap of thinking the campaign investment was worthwhile. However, would you have made 120,000 worth of sales without advertising investment? Measuring the effect of doing nothing is very important to predict ROI. Unfortunately, not many organisations have an appetite for doing nothing!
Predicting the positive and negative effects of planned campaigns is very useful for predicting the effect of an action on your customers. A Telco we worked with wanted to predict which customers were likely to churn and then contact them to offer them a plan. If doing predictive analytics properly, there is always a hold out group who are not included in the campaign, which there was in this case. They found was that people who were contacted churned at the higher rate because that contact prompted the customer to think about their plan and whether they could get a better deal elsewhere. That was a negative action from their campaign that they didn’t forsee. If they were predicting both the positive and negative affect, they could have changed the tactics of their campaign.
Marketers know the importance of communicating with the customer when and how they want. Do they like emails, SMS, phone or direct? Do they prefer to be contacted at 7pm or 2pm? For example, in the energy sector, you could predict the likelihood of a customer taking the action you want them to take (such as paying a bill) based on their predicted communication preferences. For example, if you contact them via SMS at 2pm reminding them about that bill, are they likely to act? Or do they prefer that reminder at 7pm via email?
Calculating when a person is likely to buy a product next will help personalise offers and decrease time to next purchase. You don’t want to send a customer an offer for a toaster when they have just bought one. You will want to offer them a product they are more likely to need based on data. Another scenario might be that they have purchased milk in the last five days and you want to predict when they are likely to need that product again soon. Inter purchase interval can vary significantly from person to person and household to household. Estimates and gut feelings are not enough. A data driven approach is needed.
Predicting the likelihood a prospect will buy but also the likely value of that purchase is a powerful insight to have in your marketing and sales arsenal. You can increase productivity and profitability of the Sales team by prompting them to contact leads that are likely to buy and provide the most in terms of value. For example, Sales have limited time and a choice between spending time with a lead with a 20% chance of converting and is likely to produce a profit of 100,000, versus a lead with a 50% chance of converting at a profit of 20,000. Who are they going to want to spend time with?
If you want to measure it, ultimately you will want to predict it. However, what you predict comes down to what success looks like to your team and business and what your marketing goals are.