From Obamacare to Marks & Spencer: when websites go wrong

M&S's unpopular new site is not the first to damage a firm


The woman in charge of Marks & Spencer's internet presence back in September 2012 was quite sure what was holding back online sales: its website, which had been run for it by Amazon under a deal signed in 2005. The M&S director in question, Laura Wade-Gery, said: "Amazon sells everything as if it was a book. We've tortured their system to do what we want it to do."


The retailer began building its own and in February 2014 M&S unveiled a new "magazine-style" site, which drew plaudits from some reviewers. But as well as early reports of crashes, customers apparently felt that there was too much video, that they couldn't find what they wanted, and that the site was "awkward to navigate".


So last week M&S knew what to blame for the news that online sales had fallen 8% in three months: its website. Finance director Alan Stewart bravely insisted that "It is a bit like going to the supermarket for milk, they've moved it and you can't find it immediately." But an 8% drop would never happen at an internet business such as Amazon, because it would have spotted it at 1% or 2% and rolled back to a better-performing version. So how can two years of effort by one of Britain's biggest companies produce something that fell so short?


According to people in the industry, it is probably because the process took two years. The more monolithic a computer project becomes, the less likely it is to withstand the continual shifts in the way that we use the internet. "For M&S, the fact that it took them 18 to 24 months means whoever did the project planning was two years behind the curve," says David Hamilton of Dog Digital, a media services company that has built websites for private and public sector clients. "In our world, it's about agile development – designing, failing fast, optimising. You don't go for a 'ta-daaaa!' event."


The biggest sites, such as Facebook or Amazon, are changing all the time, Hamilton points out – yet they never promote the fact, and the incremental, small nature of the changes makes them hard to spot. But that also means that if a tweak fails – which can easily be measured via statistics such as sales conversions or time spent on the site – then it can be abandoned and the site returned to how it was.


It's very often public sector projects that fail badly. Last year, the Obama administration was severely embarrassed when the political triumph of implementing healthcare for all was undermined by a website that crashed repeatedly and suffered security glitches. In the UK, the Department of Work and Pensions' efforts to implement universal credit were so catastrophic that a "tiger team" of programmers was sent in by Francis Maude's Cabinet Office to implement a version that would work for the majority of claimants, rather than trying to cover every case.


Yet private sector sites can fail badly too. The canonical example is boo.com, a would-be sports clothing retailer from the first dotcom boom of the 1990s, which built a fabulously complex site able to fulfil orders in a dozen European countries. It offered animations, 360-degree views of clothes and shoes and a virtual shop assistant. But its Swedish founders overlooked the fact that the vast majority of people had dial-up connections and comparatively slow computers – meaning the front page alone could take minutes to load. The company closed after burning through many millions of pounds.


The key problem is that the ground has shifted radically since M&S embarked on its change. In September 2012, fewer than 60% of Britons had a smartphone. Now that figure is over 70%. It's no longer just aboutonline shopping, as it was in 2012. Now it's about mobile shopping. Even among the over-55s, more than one in four now own a smartphone and 61% of those people use it to compare prices, according to Kantar researchers.


For the young buyer that M&S wants to attract, the focus has shifted from the mobile web to apps – a category that was significantly smaller in 2012. If you're out and you want to order something, you can fire up a store-specific app. While M&S does have an app for iPhone and Android, it has received poor reviews since it was updated in March, with complaints about usability.


And who's to say that in another year or two the way that people use their smartphones won't have changed again? Retailers already complain about people "showrooming" – visiting their shops, looking at items and then seeing if they can buy them cheaper online. Amazon is looking to accelerate that with a feature in its new Fire Phone that can identify an object, a picture or a snatch of video via the camera, and show you the item on Amazon – where of course you can order it instantly.


"The biggest issue most organisations have is their culture from the boardroom down," says James Stevenson, managing director at digital consultants Bletchley Group. "They specify their requirements, and then one, two or three years later feel pleased they have delivered it only six months late and only 15% over budget. Of course they miss that what has been delivered is already out of date, and no longer what their customers wish."


The internet age requires a whole new way of thinking about selling products, says Stevenson. "Too many organisations still think in 'projects', when the reality of business is that these are actually digital products, and need to be managed as such over the lifetime of the digital product."


But he sees positive signs inside M&S: "They have started a new way of working and organising themselves, into digital product teams, and are working on being customer-focused and more agile." The company's latest results are the outcome of the old ways, he says. "I'd be optimistic they will see the results of this very soon."