Retailers need chief algorithm officers to make e-commerce pay

Michael Ross, chief scientific officer at eCommera, an ecommerce software and services company, calls for chief algorithm officers

The BBC’s Today programme has a slot for company financial results that often features the CEOs of retail companies claiming that, whatever difficulties their physical stores may have, their online businesses are soaring away.

These segments are probably even less listened to than Thought for the Day on the same programme. In any case, Michael Ross (pictured), chief scientific officer at eCommera, an e-commerce software and services company, is quick to pour cold water on such protestations.

“They do often say they are going great guns online from a short-term revenue growth perspective, but they rarely say by how much, and even more rarely discuss profitability or predict how it will go over the next few years," he says.

“Customers just have more choice now, and retailers have to be a lot smarter. Amazon is simply depressing the margins of everything it touches.”

Ross has the dinner table distinction of being a Cambridge mathematician and McKinsey alumnus who yet founded a lingerie retailer online – eCommera’s client list includes many of the great and good of British retail, from Asda, through House of Fraser, to T.M.Lewin, purveyor of shirts.

More on e-commerce data analytics

He says e-retailing is very hard work, and that this is under-estimated.

“It is like retailing in the 1830s. People are just warming up to a complete rewiring of retail economics. As online becomes 20% or more of your revenue it becomes crucial. The numbers are getting big now for every retailer.

“But, the economics and profitability of retail stores is very transparent. Not so online. It is extremely complex to unravel the economics of an online business. The costs are more opaque, there are lumps of digital marketing spend that are hard to understand. And the revenue is harder to predict.”

“The past 10 years has been an executional phase: retailers with great propositions that have executed well. Growth has inevitably slowed, from five, seven years ago. The focus now needs to be on optimisation. And that is a whole different bunch of skills. What got you your first £100m online is not going to get you the next £100m,” he says.

The need for people who can ask the right questions

“We need chief algorithm officers. Algorithms – here, data plus logic – are springing up all over organisations, and they need to be understood and managed. It is increasingly difficult to understand what is going on in your online business without an integrated data analytics system. Excel is not big enough, and my layman’s definition of big data is things that are too big to go into it.

The focus now needs to be on optimisation. And that requires different skills. What got you your first £100m online is not going to get you the next £100m

Michael Ross, eCommera

“But a nice feature of the online world is that decisions are reversible. If you get decisions wrong, you can change your mind. Get the location of a physical store wrong, you are in a bad place.

“In the old world, data was hard and memory was expensive, so the scarce resource was on the technology side. But now, with this proliferation of tools, the hard bit is asking the right questions. You don’t go to your Hadoop guy to find those out. People who can ask the right questions are really hard to find. Ask a pointy head in the corner and they will ask either the wrong questions or banal ones.”

He takes a sceptical view of the current proliferation of data science institutes, including UK government interest in big data.

“There is a bandwagon here. I’m not sure if George Osborne would know an algorithm if it stared him in the face. But the main thing is you have people coming at this from Hadoop and NoSQL, for instance, and that is one part of the jigsaw. 

"But as those become more common, organisations will realise that general managers and business people are ill-equipped to use those. More relevant is statistics and decision theory, invented 100 years ago by John von Neumann.

“The skills required are actually quite old-fashioned.”

Read more on IT strategy