Testing APIs in eCommerce Sites – Surveillance Tests
Implementing quality tools under high pressure and short deadline conditions requires development and quality organizations that stand firm on the issues of process integrity. When done rigorously, this provides a resource and a path for software quality to become a practical reality in eCommerce enterprises. And, an attractive, always-functional site is the path to a customer’s heart.
However, in spite of the best software quality efforts, bugs and unforeseen interactions will continuously plague live eCommerce sites. And nowhere is this more evident than in the operation of the APIs that glue the system together. By intent, the use of eCommerce systems is as rapacious as their evolutionary design. So, when a problem does slip through the high-intensity development/test processes, it will wreak havoc in short order due to the high volume of 24/7 transactions that it is expected to support.
A derivative of the sanity check concept is the surveillance test system that directly exercises the live site. This test approaches the site from the user’s perspective, verifying paths from introductory pages to shopping cart APIs through purchase and delivery of products. The purpose of these tests is to catch live site problems before they cost the company sales and customers.
Surveillance tests are normally run continuously with cumulative statistical reports issued directly via email messages. They can be tied into the systems that measure marketing metrics because, in many cases, the intentions of the two converge. Marketing is concerned with any aspect of the site that runs counter to turning browsing into purchase conversions. This is exactly the effect of a defect in the site. The customers are put off by a malfunctioning page or feature and are discouraged from pulling out their credit cards and making purchases.
Maximum effectiveness comes from implementing tests that exercise all high visibility site features and can run in near continuous loops. Results reporting is necessarily done on an exception basis with the output displayed in as attention-getting a manner as possible. One of the more common output display techniques is direct email alerts to the engineers involved with each segment of the system. Another valuable methodology is the use of trend reporting. Watching the rate of change of monitored aspects of the site will provide an early warning of a problem while the actual number of occurrences is still relatively small.
Find out more about Web API testing.