Predictive analytics takes online marketing and ROI to the next level
No one can deny the importance of predicting the future, or its difficulty. After all, it was Microsoft’s Steve Ballmer who predicted back in 2007 that the iPhone wasn’t likely to achieve significant market traction!
At the inaugural Ecommerce Club Members lunch in February 2015, we were lucky enough to have Hedley Aylott, CEO of Summit and Dan Patmore, Search Strategy Manager at Argos, discussing the role of predictive analytics in marketing.
Accurately predicting the future performance of marketing activity is now possible thanks to the use of advanced statistical techniques called predictive analytics. Statistical modelling and the understanding of customer buying triggers can result inhighly accurate predictions of where and when to spend marketing budget for the greatest return The ability to understand customer buying triggers finally puts control back into marketers hands. With the right information, brands can identify where to spend their budget for the greatest return.
Today’s retailers are driven by margin, and the need to increase sales with a growing margin in an increasingly competitive market. What brands need to do is understand what driving consumer behaviour is and respond to it immediately. According to Aylott, that is the service that Summit’s Forecaster technology can provide. He says, “We set out on a journey to move from ‘what happened and why’. Then we recognised the key lies in understanding what happens next.” Retailers need to think about how budget is set and choose where to spend their money. In today’s ever changing and increasingly fragmented market, simply following a set of rules doesn’t work and it won’t work in the future.
Summit’s forecasting software analyses past customer behaviour and predicts which triggers are likely to impact the market, so it can tell brands what to optimise against. At a macro level, aAs Hadley points out, all ecommerce is seasonal and if brands can truly understand consumer behaviour, they’ll understand the buying triggers.
Weather is obviously an important, and predictable, trigger. In the UK, at 20 degrees the whole country strips off and many decide to buy a paddling pool. At 4 degrees snow doesn’t melt when it falls, so with a 10-day forecast of 4 degrees and 8mm of rainfall, it’s time to get the sledges out. As Aylott puts it, “Seasonality is about customer service and it’s the reality of customer-centricity.”
Aylott says, “The predictive model sets the budget for a day based on historical sales data. It outlines an optimum spend, provides a revenue forecast and tells you where to spend the money, and it does it automatically.” He says, “It’s all about behavioural psychology and the wisdom of the crowd. The trouble is that no one person has a sufficiently wide data set to make an informed decision.”
As it makes its predictions and tracks changes in behaviour, Forecaster automatically updates instructions to Google and reports to Argos. Critically, however, it is possible to override the system and that’s vital when the business may have different goals at different times. Patmore says, “Having a strong logical base enables the making of what might seem like irrational decisions, dependent on the basis of the decision, or particular business goal .” Sometimes the goal is to increase market awareness and brand positioning, sometimes it’s about balancing the most effective spend against business needs (eg is it time to get rid of seasonal stuff filling the warehouse).
What matters for the brand is understanding the relationship between budget and return. The question to ask is whether the brand goal is profitable sales, engagement or brand. Once the strategic decision is made, the technology today will facilitate that goal. It’s about understanding profitability against media. Aylott says “It’s not about big data but about intelligent data and Patmore agrees. What predictive analytics allows brands to do is get closer to the customer. He says, “That’s the critical part of search. There are an infinite number of variables but the customer lies at its heart.