<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=1005900&amp;fmt=gif">

Insights

Don’t look back in anger… how to keep your website performing

04th September 2024 by 
Danny Quilton Prepare for Peak

Peaks and troughs – that is the nature of the ecommerce business model. When the cloud infrastructure that your website (and indeed your business) is built on is not up to the challenge of a significant surge in website traffic you run the risk of tarnishing your brand reputation and making your customers very unhappy. A case in point: Ticketmaster and the Oasis ticket shenanigans over the weekend. 

Last Tuesday the British public drew in a collective breath of awe the moment Oasis – Britpop legends of the 1990s – announced not only that the brothers Gallagher were getting back together, but they would be putting on a series of concerts in London, Cardiff, Manchester, and Edinburgh for the first time in 15 years.  

With pre-sales of tickets happening on the Friday and general sales opening on Saturday, many fans were disappointed with their experience on the Ticketmaster website. Thousands waited in queues upward of 5 hours; the lucky ones got tickets, the rest were either kicked off the site once they’d reached the front of the queue, or received one of a number of error messages – Error 503, website not found, session suspended (we think you’re a bot) and not authorised to perform this operation – that didn’t allow them to complete the transaction.  

At Capacitas, in our experience of working with ecommerce businesses, we understand that optimising website performance is not always easy, especially when dealing with large-scale traffic surges. We typically see that organisations fall into one of three camps: those with no strategy to cope with scaling; those with an ineffective approach who are missing the major issues; and those with an effective approach characterised by overspending on cloud and focusing too much effort in the wrong places. 

We also understand the reasons for this – the complexity of performance and scalability; complicated legacy systems; increasing frequency and severity of peaks that is outpacing an organisation’s ability to prepare for them; and a focus on more functional aspects (i.e. new products and features) instead of non-functional aspects like performance and scalability. 

The good news is that successfully planning for peak periods can be done.  

Solving performance issues  

Ticketmaster is not alone in experiencing website performance issues. Other ticket sellers have fallen foul, while just about every Black Friday/Cyber Monday, seasonal promotion causes chaos on the websites involved.   

Why then do these sales events crash ecommerce websites?  

The easy answer is lack of technology infrastructure. Websites typically crash when there is a surge in traffic because their systems (i.e. their software or hardware) or their third-party systems, cannot cope with what is being asked of them.  

 The longer answer is a little more involved. 

Understanding your capabilities & capacity  

Yes, businesses can harness additional cloud computing capacity to cope with the additional demand. However, while cloud is essential in preparing your system to cope with peaks, it is not the be all and end all solution to your scaling challenges.  

Ecommerce organisations have complex IT estates comprising legacy systems that need re-engineering in order to work in the cloud. In addition, the ‘shop front’ of ecommerce organisations, the website, is also a multifaceted ecosystem of components that must work in harmony – from firewalls and load-balancers to spread out demand, to the web servers that handle the actual traffic, application servers that handle business logic and significant database servers. And that is not to mention other elements such as the payment gateway, or stock management and logistics systems for businesses that sell a physical product.  

Adding a bit of additional cloud capacity to that already complex ecosystem without understanding how it all works together or understanding how it will perform at scale will not solve the issue. And it is important that all components in the ecosystem can scale to make use of the scalability of the cloud. 

A 10-step process to prepare for peak 

Your website, and the systems that underpin them, needs to be fast, scalable, and robust. But how to achieve those three things? Ultimately, it is about ensuring that your cloud estate is optimised in terms of both cost and capacity to deliver the performance, scalability, and reliability you need, regardless of whether you are operating in a peak or a trough.  

All ecommerce businesses, but especially those like Ticketmaster where huge traffic peaks are par for the course, need a strategy to deal with increased demand.  

So, where to start?  

Step 1 – Start preparing early 

Peaks can be caused any number of things – marketing promotions, launches, a viral video on TikTok, competitions or seasonal events – so knowing when they are happening is the first step to bolstering your readiness.  

Step 2 – Set targets for success  

You need to define what success looks like. This should be more than hitting sales targets because even if you do but your customers are experiencing slow page loads or being placed in a queue (in the case of Ticketmaster, some fans were as far back in the queue as 500,000th). One example is: X% of the time, during the peak period, the page load time of key website pages should be Y seconds or less. You should set targets across all 7 pillars of performance

Step 3 – Forecast peak demand 

Next, you need to understand exactly what will be asked of your website; and create a target demand figure to work towards so you can plan the website’s capacity based on this demand. There needs to be a balance here – set demand too high and you run the risk of spending too much on capacity. Whereas if you set the targets too low, you run the risk of insufficient capacity and as a result a website crash. 

Step 4 – Eliminate risk from 3rd-party systems 

As mentioned earlier, websites are comprised of a number of different systems, including third-party ones that deal with things like payment, delivery, etc. While these systems are beyond your control, the consequences of if they crash will still have an impact on your business. As a result, assess the risk, build resilience, and implement governance processes and SLAs to manage / negate the risk. 

Step 5 – Create a contingency plan 

A contingency plan will help in you prepare. You need to understand what risks you face during a peak period; how critical they are and how you are going to deal with them. An example of a risk is what if the site demand exceeds what we have forecast. An example of dealing with this risk would be to turn off non-critical website features to preserve site stability. 

Step 6 – Test, test, test 

As the title suggests, testing your peak plan is the only way to ensure it will work and understand how your systems will react in different scenarios. This is a crucial step in ironing out problems and ensuring you can meet your forecast demand. 

Step 7 – Deliver ongoing assessments of readiness for peak 

The ecommerce world doesn’tdoes not stand still and neither will your business. In the months leading up to peak period things will change on your website independent of your planning. These changes need to be factored into the planning process and the impact to peak readiness assessed – using a consistent methodology to analyse the data and deliver insights. 

Step 8 – Ensure your post-sales systems are on the same page 

It is not just your website that will be working overtime during peak periods; your other systems also need to be able to cope with the increased demand, such as delivery, returns and customer support. These systems should also be analysed in terms of risk, scalability, and performance, and tested accordingly. 

Step 9 – Create a peak playbook 

This step will help your team deal with the peak period once it has arrived. The playbook should cover the people, processes and tooling needed to support the period. It should include information such as duration of the promotion, timings of advertising campaigns, who is responsible for what (roles and responsibilities), monitoring processes, procedures to implement contingency plans, etc. 

Step 10 – Conduct post-peak review 

Once the peak is over, it is important to analyse what worked and what did not, and using those insights to improve your approach to deal with forthcoming peaks 

What’s next? 

This is, of course, just a high-level answer; a good place to start preparing your ecommerce business to deal with peak periods successfully. 

 


 

If you would like more detail on our 10-step approach, you can read our guide. 

Or book a session with one of our experts to discuss how we can help your ecommerce business create and maintain a successful preparing for peak strategy. 

 

  • There are no suggestions because the search field is empty.
Filter by Tags:
SRE
AWS
cto
ITV
TSD
cfo
cio