Beating the Black Friday blow out
Were you a part of the online shopping frenzy on Black Friday? I was and I know many consumers suffered problems with websites going down or being unavailable. By way of inspiration, I thought I would take a look at some industry research which showcased website availability (or lack of) in the UK and then highlight a few clear steps to demonstrate how retailers can avoid a website going down during major retail events.
Black Friday was widely reported on this year as shoppers opted for online over in store shopping. As Britons spent 1.1 billion pounds shopping online, an increase of 36 percent on the previous year, unprepared websites suffered.
However, according to TrafficDefender’s live leaderboard, between midnight and 6am on Black Friday a number of online retailers had already experienced issues with availability.
The web traffic management and queuing system expert watched the likes of Schuh and Pretty Little Thing experience periods of unavailability in the early, pre-office hours, but they quickly recovered. By mid-morning technical difficulties had settled down and most site’s traffic levels evened out.
As expected, traffic levels peaked again around lunchtime and into early evening, affecting even top online retailers such as John Lewis. River Island also got into a bit of trouble immediately after its deals went live due to the unprecedented volumes of website traffic.
So when even the experts can’t predict the impact of Black Friday, how does a website prepare for traffic levels like this?
It’s all about scalability. With most online retailers living in the fear of a repeat episode of Black Friday 2014, most had put in the preparation ahead of time.
In contrast to previous years, TrafficDefender monitored a consistently high level of traffic throughout the day rather than massive spikes, indicating that marketing and IT departments were working more closely together to flatten out demand.
Whilst it may seem like an obvious thing to do, many online retailers invested heavily in scaling up their infrastructure, however, it’s still won’t help to avoid traffic driven outages.
Unless a website or application is built for the cloud then scaling infrastructure typically delivers proportionally diminishing returns in capacity, plus it’s totally impossible to accurately estimate the size of the infrastructure required in events like Black Friday when there’s no way to know the level of traffic that’s going to hit.
Experts in this area talk about ‘Performance by Design’ which is the idea that to achieve fast, robust and reliable systems, performance must be built into websites from the earliest stage. According performance experts Intechnica, what successful online retailers need to do to prevent slow downs and outages in these peak times is to build a website specifically for the cloud, with the infrastructure to take on large amounts of traffic. However, the cost to a business of putting this extra infrastructure in place can make it hard to be profitable during sale periods when margins are already tight.
This year, as an alternative, retailers such as Tesco, and Argos were able to stay online and trading thanks to the use of a queue. Queuing works alongside a website’s existing capacity management to divert any excess visitors to a first in, first out queue whilst the site in question deals with as many visitors as it is actually capable of handling. Clearly infinitely scalable infrastructure would be the ideal solution but the sheer cost of this makes it price prohibitive. Queuing therefore is a very sensible solution and something that consumers seem more than happy to do, given the circumstances. This ensured that certain websites continued to function even when the volume of traffic was exceeding the levels that the infrastructure can normally handle.
It’s not all bad news for those sites that went down. Clearly those companies who suffered website issues on Black Friday have a good marketing strategy in place and the surge in traffic is the sign of a good retailer. It’s a nice problem to have, but measures can be taken to mitigate this kind of risk for the future.