The retail sector has always been data rich yet insight poor, in relative terms. There’s a lot of data in retail, from the point of sale systems, through ecommerce systems, through social media, from customer contact centres, DCs, and so on. But building meaningful insights on the back of all that is not exactly simple. We want to hear your thoughts, opinions and observations on what data can do for the retail operations and logistics industry.
eDelivery.net’s editor, Sean Fleming, caught up with Matt Hopkins, retail director at Blue Yonder – a predictive analytics specialist – to hear about machine learning algorithms, and how to get valuable insights out of mountains of data.
Most retailers across all segments of retail, groceries, fashion, and so on, are struggling with the fact that they basically now operate at the clock speed of the new customer. We know the new customer is very fickle, they shop on price, they’re very difficult to predict in terms of how you manage certainly the supply chain and even the physical image in store.
So really the context is, where new customers are concerned, the new normal is that they’re very fickle, which causes challenges in things like multi-channel where a lot of retailers are struggling with their inventory, the cost of goods sold, which is obviously the supply chain dimension, and they’re trying to figure out how to align their supply chain and synchronise the data within that supply chain from the customer right the way through to store through channel, right the way up through to warehouse management, and out in the street.
The biggest contribution data can make to all of that really is deliver insights and a level of transparency .
There’s a complexity now which is, how do you match investments being made around technology in-store, and certainly in ecommerce, we’ve seen a lot of retailers who’ve invested very heavily in building these new channels. They haven’t necessarily looked at the economics of retailing across a longer channel structure, and they certainly haven’t looked at putting investment into the supply chain in terms of matching that pace that the customer operates at.
So what I see as one of the biggest challenges really, is how you operate at scale, how do you take those insights that are coming from your data, and deploy them so you can actually start reducing the amount of stock, the number of write offs and mark downs, start accounting for things like holidays, seasonal effects, any kind of variables like store holiday.
The answer, in large part, is data. But it’s not just data. It’s the digital applications driving that data into the operation, so that is pick, pack, shift, move, buy, plan, source, whatever it’s going to be.
There’s a huge amount of investment in analytics, taking place. But if you’re doing that across a store estate or 3,000 stores which carry 40,000 products, you have 20,000 people in the store serving x million customers, the analytics then sort of become, well, less than useful. You have all this data, all the analytics, but how do you actually react and turn them into actionable insights.
The actual insight, in my opinion, is in predictive applications, which are a combination of big data, and machine learning algorithms. Those two things become basically habitual applications, and that allows you then to automate and respond to what’s going on in your retail ecosystem.
So a lot of retailers now are using analytics quite well, but it doesn’t solve the problem of operating at customer speed because you can’t get those insights into things like personal parameters without using hundreds of people to actually put in parameters and set rules. The KPIs instead are very important and those KPIs are, what’s the cost to serve, how much inventory do I want, how do I want to trade off if I’m a grocer, my write offs from fresh produce versus sending another dispatch, so those particular applications are really the next step for a lot of retailers in actually beginning to use the data in a really effective way, and actually which is more relevant to the way they operate their business.
I think there’s a huge opportunity where machine learning algorithms are concerned. I think Gartner actually said this year, 2016 was going to be the year of the algorithm. So that should get everyone excited, I guess.
So take an an example a new store or a new product. Machine learning algorithms are far better at looking at the variables, until you start picking up a sales pattern. Things like new products and new stores are big complex headaches for large retailers, and the machine learning algorithms allows you to adjust very quickly without a reliance on manual intervention. But they also allow you to factor in what you know as well about your business, about the customer, the weather, the store location, all at a key store level.
Machine learning is going to be hugely impactful in terms of getting the economics of your channel right, of how you want to serve the new customer, create the right customer experience in store, and without having machine learning, retail’s then really not going to be able to adjust the supply chain and calibrate it properly.
In terms of the delivery side, if you’ve got a very accurate fiscal baseline, from demand forecasting to having a very accurate view of what you pick, pack and ship, that is going to give you a very high standard and quality of supply chain forecast. Most retailers now have multiple forecasts, you’ll have a warehouse inbound forecast, you’ll have an outbound forecast, you’ll have lots of forecasts, and that creates buckets of inventory and it becomes
The better the forecast is, the more you can calibrate your delivery services around that. With returns, it’s about understanding the impacts of things like new products in your range and looking at like-for-like behaviour.
So getting the items right, definitely getting the right insights to drive those positions from a sourcing and buying point of view, will definitely reduce the amount of returns, and we’ve seen that with a number of our customers.
If you have thoughts on the use of data in the retail operations and logistics industry, why not get in touch? You can leave a comment, or drop us an email. We’ll be publishing some reader opinions on this topic, so add your voice to the conversation.