The majority of shopping carts provide consumers with an unpredictable and unsatisfactory experience, making Web shopping a lottery. E-consumers are prevented from making purchases on UK web sites for 9 hours and 30 minutes a month on average, (115 hours a year). 80 per cent of web sites perform inconsistently with widely varying response times, timeouts and errors – leaving consumers at best wondering what to do next and at worst unable to complete their purchase successfully. This is potentially costing e-retailers millions in lost sales from consumer frustrations.
These were the key findings of the Ecommerce Performance Study 2004, by web testing specialist SciVisum, which carried out 24/7 monitoring of the online buying process on UK retail web sites over a period of one month.
On average UK shopping carts in the study were out of action for more than ten times the accepted industry standard. 1 in 5 carts did not function for 12 or more hours a month and over three quarters failed to meet the standard service level of availability of 99.9 per cent equating to a maximum 43 minutes of accumulated failures a month. A leading DIY chain had shopping carts that failed to work for more than four days during the study, making it the worst performer.
“20 million UK shoppers are expected to spend £17 billion online this year. Many UK e-retailers have made significant strides in improving consumer experience, and we are not surprised that most of the best are IMRG Members. But the appalling state of other retailers’ web services is both shocking and completely unacceptable; it blights our industry, and we are grateful to SciVisum for revealing it. Their study shows that bad website performance will cost merchants at least £225 million in 2004 if they don’t get their act together,” commented James Roper, CEO, IMRG (Interactive Media In Retail group).
The SciVisum Ecommerce Performance Study 2004 tested the online buying process, specifically the ‘add to cart/basket’ function, of the web sites of over 50 of the UK’s leading e-retailers for a period of four weeks during April 2004. It also identified daily and weekly trends in online shopping and industry wide peak shopping times for e-consumers.
“UK ecommerce sites are slapping customers in the face, rather than shaking them by the hand. Turning consumers away once they have made the decision to buy is commercial suicide. Although specific steps of a transaction may fail to complete, in most cases the web site itself is still functioning, so it is likely that many online managers are completely unaware of the problems,” said Deri Jones, CEO, SciVisum.“Web site operators need to stress test the crucial functionality on their web sites, down to the transactional level to assess the full scale of the problem. On average the ‘add to cart/basket’ step failed to function correctly in more than 1 in 100 cases. This is just the tip of the iceberg, since consumers typically buy 2/3 items at a time and perform multiple steps to complete a purchase, the true number of consumers let down will be more than 1 in 20.”
Best and worst performers
UK web sites displaying the worst errors included a leading high street mobile phone store, a well-known music/DVD retailer and a leading DIY chain, which accumulated more than four days a month without shopping carts functioning.
Only 20 per cent of shopping carts were able to handle daily and weekly traffic patterns consistently, and the best performers included John Lewis (www.johnlewis.com), Waitrose Direct Wines (www.waitrosedirect.co.uk), Orange (shop.orange.co.uk), Figleaves (www.figleaves.com) and WHSmith (www.whsmith.co.uk).
“In ecommerce every transaction counts. It’s the sporadic and unpredictable performance of web sites that is the most worrying to users. Our study shows that web sites are not coping with the heavier loads that occur day by day, failing through inconsistency to support consumer-buying habits and completely missing the boat when it comes giving e-consumers a satisfactory, reliable experience. Online retailers need to test and monitor crucial site functionality or they will lose customers to more nimble rivals,” added Jones.
The 51 leading UK ecommerce web sites tested included retailers of music, books and videos (18 per cent), clothing and fashion (23 per cent), catalogue department stores (22 per cent), consumables and electrical equipment (14 per cent). The remaining 23 per cent of web sites tested consisted of web sites selling DIY products, mobile phones, cameras and gifts.
The SciVisum Ecommerce Performance Study 2004 tested the ‘add to cart/basket’ function of over 50 of the UK’s leading e-retailers every 10 minutes for a period of four weeks during April 2004.
Two measures were used to profile this functionality – HTTP errors and response time. These highlight the two main failure areas: HTTP errors, which make it impossible for an e-consumer to complete a purchase; and response times greater than 30 seconds for HTML download, causing the majority of e-consumers to abandon the transaction or assume an error has occurred. The entire purchase process was tracked but only the last step of the transaction process was profiled in the study.
HTTP errors (causing more serious problems for consumers) rather than timeouts were the major cause for shopping cart failure. This contrasts with the misconception that slow page delivery and timeouts occur more commonly than HTTP failures.
Based on the findings, SciVisum made a number of broad recommendations to improve the ‘add to cart/basket’ function of web sites:
1. Review the ‘add to cart’ function of your website to ensure that the server load is kept to a minimum. For example firstly strip any HTTP ‘POST’ data down so that only essential variables are passed within it, such as the product part number; and secondly avoid changing cookies at this crucial stage of the purchase process.
2. Analyse web systems for ‘database locking’ flaws, which can produce errors at load levels well below the capacity of the server hardware; and require ‘high simultaneity’ testing of each function to be identified.
3. Be aware that although ‘add to cart’ functions may perform well in ‘once off’ or ‘normal use’ testing, only simulated-user load/stress testing of the functionality will expose underlying problems that cause more sporadic failures; even 1% failure during busy periods is 10 times too high.
4. Implement a regular monitoring and test programme that exercises complex functionality and multi-step user transactions as well as basic pages over time – to help identify as soon as possible when the system is struggling. Symptoms will include increasing failure percentages although page-average measures may be steady.