SciVisum, a web-testing specialist, made a number of broad recommendations for e-tailers to improve their performance:
1. Simple uptime/downtime monitoring of your home page and/or a few main pages simply won't reveal how the shopping transactions are behaving - 24/7 functional monitoring, running multi-page User Journeys that mimic real users' product finding and purchasing transactions on-line is what is required. Ensure that these KPI metrics are measured throughout. For the eCommerce engineers:
2. Review key transactions such as the 'add to cart' function of your website - to ensure that the server and database load is kept to a minimum. Firstly watch out for HTTP 'POST' data bloat, so that only essential variables are passed within it, such as the product part number; and secondly avoid adding or changing cookies and sessionIDs during the crucial later stages of the purchase process.
3. Analyse web systems for 'database locking' type flaws, (e.g. is there is a limit on how many users can concurrently add a database line representing their purchases) which can confusingly produce errors at load levels well below the capacity of the server hardware, which makes it hard for the IT team to identify the problem.
4. 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 higher than 99.9% Service Level Agreement requires.
5. Whether managed in-house or out-sourced, your web site is likely evolving and changing all the time, to respond to marketing demands, and to add to capacity and performance. These changes often cause inadvertent decline in user experience and transactional effectiveness. Thus the IT and marketing team managers should agree an ongoing program of testing and monitoring, to allow evidence-based decision making on future upgrades; the test regime should include 24/7 functional monitoring, regular stress tests, perhaps twice-yearly, and ad hoc trouble-shooting audits say yearly to ensure the overall design and infrastructure is not losing its edge.