Member Article

Prime Time to Prepare: what we can learn from Amazon’s ‘Prime Day’ struggles

One of the big contenders to Black Friday and Cyber Monday is almost upon us once again. The 11th July sees the return of Amazon’s ‘Prime Day’, the third of its kind. Starting at 6pm on Monday 10th July, and running for a bumper 30-hour period, the retail giant makes a range of special deals available exclusively to its premium membership, and encourages users to sign up to their Prime service. Looking at year on year sales growth, ‘Prime Day’ generates hundreds of millions of dollars for Amazon’s top line and is one of their flagship events. So far, so good.

Yet even Amazon, one of the largest and by far the most advanced of all e-commerce retailers, has experienced website problems with their Prime Day promotion. In 2016 many unhappy users took to Twitter to complain about basket issues and long checkout times, with the hashtag #PrimeDayFail. Profits increased from 2015, but not to the levels that Amazon themselves expected – they were down hundreds of millions of dollars on the ‘expected value’ of Prime Day. The cause of the strife? Many experts observed that ‘unexpectedly high’ levels of user traffic could have caught Amazon unaware and led to performance degradation.

When user experience fails to live up to expectations, the knock-on effect for companies is significant. Amazon may point to the fact that Prime Day was their largest retail day ever, but profits fell well short of pre-event hype levels. In the run up this time last year, some experts were expecting a $1 billion dollar ‘day’ – but the final figures weren’t really close.

If the best can fail, what can other ‘mere mortal’ companies learn about preparing their websites for peak traffic events? Here are three key steps to keep organisations of all sizes performing at their best:

Anticipate: for any site preparing for a regular largescale event, there’s no excuse for failing to anticipate user figures. With Prime Day now an annual event, by now Amazon will have a good understanding of potential peak traffic. Similarly, sites that offer any regular event (a sale, blog post, exclusive interview etc.) should be learning from past stats to predict the future of what they need to be expecting.

Test: to make sure that your site can cope with expected visitor numbers, invest in load testing and concurrency testing models. These allow insight into not just how many total users your site can handle before it fails, but also site performance during high numbers of concurrent users. This could be everyone trying to grab that discount TV when Amazon release a new deal, or fans rushing to buy tickets for an upcoming concert. Whatever the reason for peak traffic, it’s unacceptable for companies to slow down at any stage and alienate potential customers – testing can alleviate this.

Prepare: preparation for massive events is two-fold. Firstly, sites need to ensure that they have capacity to deal with users on their servers. Secondly, all companies need a strategy to deal with any potential issues that do crop up. A 3rd-party hosted incident page is the best way to guarantee customers can clearly understand what is going on in case of emergency outages. Testing a site’s ability to restart under critical load is another facet of performance vital to crisis management. No-one wants to suffer an outage, but bouncing back quickly will restore some customer faith and minimise losses.

When it comes to dealing with high peak traffic, failing to prepare will undoubtedly lead to missed sales and a negative perception of the brand. If Amazon’s Prime Day doesn’t proceed flawlessly in 2017, they are likely to face a backlash from social media, and even in their share price. Not even Amazon is immune to downtime or bad performance, and the same is true for companies at all levels of maturity and size. Just remember; customers’ time is money, and you shouldn’t waste it.

Sven Hammar, CSO at Apica

This was posted in Bdaily's Members' News section by Sven Hammar .