Holiday 2019: AD&D Leaders Should Test Customer-Facing Applications Early To Guarantee Performance During Peak Trading
This blog post is part of Forrester’s Holiday 2019 retail series.
I don’t know about you, but the holiday season still always finds me personally unprepared, with a huge list of gifts still to find well into December. There’s less excuse for retail AD&D professionals to be so forgetful. This blog offers retail AD&D leaders simple, actionable advice to ensure successful holiday trading.
Look to continuous testing vendors, and budget for year-end volume and performance testing. Trading volume peaks during the holiday season. You must confirm that your customer-facing applications are ready. You need to test together all the improvements you’ve made to them during the year. Work with vendors such as Cigniti Technologies or Tech Mahindra to undertake continuous testing: For information on the leading vendors, read “The Forrester Wave™: Continuous Testing Service Providers, Q3 2017.”
Mature retailers, with 10 months of their digital transformation journey complete, spend about 20% of their IT budget on testing; less mature retailers spend 30%. Why? Mature retailers exploit cloud mobile architecture, which means they can spread the testing and volume performance engineering through the year. In addition, retailers reserve between 5% and 7% of their testing budget for volume and performance testing.
Expect to spend more on testing client/server applications. Retailers with client/server applications need a comprehensive volume-testing dress rehearsal before Black Friday. For store systems, you must add another 2% to your volume and performance testing budget to establish a test environment that simulates POS hardware and firmware. To reach the right volume, you need to add another 1% for test automation. In total, this will take 8% to 10% of your IT testing spend.
Recognize that you might not be able to test everything. To avoid complex simulation, conduct limited volume testing in the lab, then move to one pilot store, monitoring resource constraints in expectation of early failure. Make fixes until the single store works at scale, then roll this out in phases to multiple stores.
I’m looking forward to hearing what you think.