Information Age (IA): Asda launched its online shopping service later than its competitors and went through a considerable period of piloting the project before it went live. Why did you choose to take this ‘wait and see' approach, and what lessons did you learn from it?
Iain Spence (IS): We launched our home shopping business in London in 1998, but were only taking orders over the telephone. So we got to home shopping pretty early but it was still very much a trial and was not something the business was really committed to. We built a system with IBM that combined a call centre and an order-taking system with workflow technology to handle fulfilment from a series of depots. In 1999, we built a CD-ROM application that would synchronise orders from customers' PCs with a central system.
Come 2000, it was clear that in order to become cost-efficient it would be better to take more orders electronically – get away from the telephone and get onto the Internet. We launched Asda.com in November 2000. We've still got some teething problems, but I think this is pretty much the norm in large ecommerce sites where there is a lot of integration required.
The big challenge for us this year will be to migrate all the systems onto one architecture. We wanted to get the basics in place so we'd have something scalable and robust that performs. That is where the focus has been, rather than on ‘bells and whistles'.
IA: How important is it for an online supermarket to personalise its interaction with its customers? Is personalisation a core feature of the site?
IS: Personalisation for us is the future rather than the past. What our customers want is something that is fast to use and simple. The last thing you want to do is use technology for technology's sake and come up with something that's going to scare the customers. So we have personalisation but it is very basic. Over the next 12 months and beyond we'll see more.
What we have at the moment you could do with most technologies. We're cookie-enabled, so we know who customers are when they return to the site and we can speed log-in. When they get into the site, they can bring up their purchase history and favourites lists.
This is the biggest thing for online grocery shoppers – the ability to shop from a list of products they've bought before. The first shop takes them around 40 minutes, but when they return it only takes about ten minutes.
IA: You plan to increase the amount of personalisation you do with customers. With this in mind, why did you decide to scrap your customer loyalty scheme in 1999?
IS: We're not afraid to be different. Tesco and Sainsbury's have plunged something like 1% of their margins into building loyalty schemes. Our philosophy was about offering low prices. We trialled it for a couple of years, but we decided it wasn't going to actually create any more loyalty or profit, so we pulled it.
What we saw and, what I'm sure our competitors saw, was that when they launched a loyalty card there was an immediate uplift in sales. On an annual basis it pales into insignificance. In our view we'd rather be 10% to 15% cheaper on price.
IA: Given that your architecture is still very mixed, what challenges have you faced in integrating the online home shopping channel – for example customer account details, delivery logistics – with your existing phone and ‘bricks and mortar' channels?
IS: The big challenge with having our new and old architectures side-by-side has been synchronising data. Before we went live, we were in a position where if customers rang up our call centre to register they'd be registered on that system but not on the web. So when they shopped online, they wouldn't be visible. So when we went live with the site we had to build an interface between the two systems.
Another problem was with PIN numbers. If customers shop over the web, they need a password to get in, but if you shop over the phone they do not. So we've suddenly got tens of thousands of customers that have been used to using the phone service and now want to migrate to the web, but don't have a password. So we have some work on the data and communications side to get that all tied up.
Because we've had the telephone channel, a CD-ROM and the web – and we're also now planning to offer home shopping through digital TV in the first quarter of 2002 – we've been trying to ensure the business logic is all in one place, with presentation layers for each of the channels. In practice, though, it's easier said than done. We're probably 70% to 80% of the way towards the purist's view of a multi-channel architecture.
IA: One of the recurring problems organisations face when introducing personalisation mechanisms is that they don't have accurate data, or have many copies of the same data, or maybe even conflicting data on individuals. How have you tackled this?
IS: When we launched the web site we had a special offer that allowed customers to get free delivery on their first order and we found we had huge numbers of people registering more than once – maybe putting in a slightly different name or different initial. So one of the things we've done to get rid of that has been to raise the free delivery to orders of more than £100. From a technology perspective, we have an address management system based on postcodes to keep address information accurate, and once these have been entered into the system they can't be corrupted. And when customers register on the web site, we check to see whether their email address is unique, which has helped. When we do outbound communications with customers we clean our data ourselves. It's not rocket science.
IA: Have your data protection policies had to change to accommodate more personal interaction with customers, and if so, how?
IA: So what kind of data do you collect on customers, and how do you get the business benefit out of it? We obviously try to track customers' order histories – that's the main thing. We have different views of our data. We have customer views so we can tell if customers have lapsed and target a campaign at them. We also have product views, so we know what items are selling well and what aren't, and what ‘aisles' and ‘shelves' within the web site are popular. We can also track which promotions are doing well. It's all within the data structures of the BroadVision ecommerce platform we use, and then we use a product from Hummingbird for the analysis. One of the things we have found – something of a revelation that almost justifies the home shopping investment on its own – is that we've now got customers placing orders for precise products. If they try to pick those products and they're not there, we know that that product was also not available in the real shops. That's information that we've never had before. If we didn't have the home shopping data to show that a particular product was short, we wouldn't know the extent of the number of customers that actually wanted to buy that product. It's a big win for us.
IA: Finally, one of the reasons you chose the BroadVision platform was that your parent company, Wal-Mart, had already installed it. How much influence does your US parent have over technology decisions at Asda?
IA: When we opted for the BroadVision platform, we were looking for something that would speed up the development process. We wanted to build something very quickly and robustly. Then, at the same time, we were acquired by Wal-Mart and they'd already bought a global BroadVision licence, so it was a bit of a no-brainer. But it was a commercial decision at the end of the day, and we've no appetite to change it.
All our applications are developed in the UK but are hosted in the US in Wal-Mart's huge ecommerce data centres in Arkansas. The Wal-Mart guys provide us with all the operational and hardware support, and take care of the middleware. But that's all about to change. We've recently made the decision that all the actual coding of the applications will take place in the US so, from now on, Wal-Mart will play a very big part. This will be a boost for us because a lot of the integration work will now de done with their support.