Retail sites fail availability test

Retail sites fail availability test


High-profile companies figure among those that miss industry target of 99.999 per cent uptime

Forty per cent of web sites operated by the UK’s largest retailers have experienced problems with availability, research shows.

During a month-long test conducted by monitoring firm MyWebAlert, sites run by retailers including HMV, Next, Sainsbury’s and WHSmith failed to achieve industry availability targets of 99.999 per cent.

Previous research by MyWebAlert into the online availability of web sites of FTSE 1000 organisations found that 43 per cent recorded significant failures during the course of December 2005, and the retail sector has fared nearly as badly.

John Earley, managing director of MyWebAlert, says the results are surprising.
‘If there is going to be a sector that should be naturally inclined towards online resilience, you would expect it to be the retail sector,’ he said.

‘WHSmith and HMV both complain that Amazon has robbed them of business, but Amazon performed well during the test period, and it is companies with high availability who will win in the e-economy.’

HMV says a site upgrade had affected availability.

‘Over the past few months we have been working really hard to significantly upgrade the HMV.co.uk site, which launched at the weekend,’ said HMV’s head of internet Gideon Lask.

‘This may have resulted in brief outages on a couple of occasions, but we purposefully sought to minimise impact on customers by carrying out work at relatively low usage times.’

The study recorded an 11-minute failure on British Airways’ site on one afternoon.

‘We have internal targets of 100 per cent between 8am and 1am. Most days it hits that on the selling and servicing side for www.ba.com, but we do have occasional glitches,’ said a BA spokesman.

The test also highlighted success stories, including O2, Amazon, Top Shop, Borders, Tesco, Debenhams and Waitrose.

MyWebAlert monitored 73 web sites between 30 January and 28 February this year, using three separately located monitoring systems. It polled each site in five-minute sequences, only alerting on the failure of two successive polls.