Christmas is coming, are you ready?

For many retailers, planning for Christmas started on Boxing Day last year. I know it sounds crazy but when you’re dependent on overseas manufacturers, long lead times for certain products and a highly competitive market fighting for a finite stock holding, it pays to be uber organised.

When you get stuck into the daily routines of managing a website, It’s amazing how quickly time passes. In the heat of battle, you rarely find the time to step back and take stock of what just happened during peak trading. If you leave it too late to do the analysis, you won’t have time to react and make your site battle ready for the next peak trading season.

This blog discusses three key activities you can and should be thinking about now and incorporating into your digital planning calendar. It follows on from my post last year discussing ecommerce trading tips for Christmas.

Goal – identify weak points from last year and create a plan to address them before your peak trading times + identify what worked really well and agree if/how you can replicate and roll-out.

Site performance (hardware and software)

How did your servers cope with sudden spikes? Did you experience any outages? If so, what caused them? It’s dangerous to assume temporary outages aren’t problems to address; perhaps they were caused by issues outside your control but make sure you validate this before moving on. It’s also dangerous not to dig deep enough to make sure you understand the root causes of current problems; don’t always take things at face value.

For example, one retailer I worked with kept getting site down messages from the monitoring tool. The stock response from the systems integrator was that the current servers needed upgrading (either dual core to quad core, or adding more servers into the cluster) but a closer analysis of the traffic volumes and bandwidth usage proved that it never spiked over the tolerances that were agreed in the SLA.

We pushed back and insisted they do a proper diagnosis. It turns out that the issue was the database. Some tables had been collecting large volumes of data on a daily basis but were never archived, so the build up of data had eventually exhausted the memory and locked the database server. Instead of investing in unnecessary hardware, a simple process was put in place to clean and archive the relevant tables daily. Result = £ saving.

Key learning

Push for a thorough analysis of performance issues you experienced during peak trading. Before committing to spend on kit, make sure the diagnosis is accurate and reliable. Learn to question what you’re being told and make sure there is data to support any hypotheses.

If you didn’t experience any problems and are expecting a significant volume increase this year, you should load test your servers and find the tipping point so you can simulate the impact of traffic spikes. If there are problems, work with your IT team to make improvements e.g. adding memory, clearing down redundant data, making sure you can spin-up additional virtual servers quickly when required.

Page load speed

Slow page load compromises user experience and also sends a negative signal to search engines. This is even more critical on mobile where Google has strongly hinted that page speed is a core part of its algorithm (openly stating the benchmark that pages on mobile should load within 1 second).

If you have any technical issues impacting page load, now is the time to address them. For example, on mobile are you deferring non-essential JavaScript? Are you using lazy loading so you don’t bloat the page on first load? Have you tuned your platform and hosting to make the first byte lightening fast?

Your aim should be to minimise the number of server calls and reduce the overall page weight without compromising the UX. In other words, don’t make the page load super fast and look awful – if users can’t interact with the page effectively, those speed gains will count for nothing.

A few things to consider:

  • Are you using compression e.g. Gzip?
  • Are you using browser caching?
  • Are you minimising assets like CSS and JavaScript?
  • Are you deferring the loading on non-essential assets?
  • Have you optimised all images for web?
  • Are you serving mobile optimised images to mobile devices?
  • Are you using CSS sprites?

Key learning

Set-up page speed monitoring tools and perform regular checks on your pages. If you detect any sustained patterns of speed decreases, escalate to your development team for investigation and fixing.

UX

You may have heard, mobile is all the rage…

I won’t quote the stats on mobile growth (see useful resources at the end for a few links) because there are so many; the key learning is that mobile has become an integral part of the consumer research and buying journey. This demands close attention to the UX across devices, providing a joined-up browsing experience.

For example, if a customer creates a basket on your desktop site, when they launch your iPhone app, is the basket recognised or do they have to start from scratch?

You may think why is this specific to peak trading? In my experience, many retailers struggle to get new tasks into an already bursting development schedule. If you leave it too late, you’ll not make the pre-Christmas code freeze cut-off. I’m not talking minor tweaks like adding a button to a page but major UX/UI design overhauls, such as changing the UI pattern for how forms work across the site or making changes to the checkout.

For example, a retailer I know is migrating to a responsive code base to help them improve the UX on mobile devices, introducing break-points and use some simple adaptive web design techniques to customise content & features based on device specific use cases. This is a nine month project that will come to fruition (hopefully) in September so the new site can be stress tested in the wild and stable prior to peak.

Key learning

Review the UX of your site on desktop, tablet & mobile. If mobile traffic is growing (highly likely), you don’t want to alienate a core section of your audience.  Plan ahead now and get tasks into the development queue where there are UX issues that need to be resolved.

Make sure you test your site and content campaigns on key devices (if your mobile traffic is dominated by iOS, test on iPhone & iPad). Use analytics data to know which devices are most popular with your customers.

Comments and questions

I appreciate that this is a high-level view and there are many, many more elements to a post-Christmas review and planning project, so I’d be happy to open the discussion to other areas.

Please drop by with comments, questions and suggestions. Please share this blog with anyone who you think would find it interesting/have something to add.

For more information on preparing for Christmas, download our free guide to preparing for the holiday season.


  • Some great points James.

    Interestingly, even the largest on-line retailers can get it wrong. It’s not uncommon for big retail websites to max-out during peak trading, which begs the question – what were they (not) expecting?

    In this increasingly competitive on-line climate, retailers need to use the best tools out there to run scenarios and predict traffic behaviours. The ultimate goal should be to deploy assets ahead of when they are actually needed – not ‘as’ they are needed.

    For the average SME ecommerce business, content delivery networks (CDN) should also be considered as a good option to improve load-times and bandwidth – especially if selling overseas.

    • Hi Marcus,

      Thanks for your comment and observations.

      Yep you’re right, there have been some high profile site glitches from well known retailers.

      I do like CDNs as they move some of the load burden away from your servers, which can really help in peak.

      We had a recent #EcomChat on CDNs, some interesting comments, though surprisingly not everyone was convinced a CDN was needed + some uncertainty over which service provider is the most reliable. In your experience, which CDNs have proven value for money?

      Thanks
      james

  • Hi James – I’ll check out the CDN conversation on #EcomChat.

    Value for money is a hard one to assess – there are quite a few factors to take into consideration. A/Bing would be the most conclusive way of assessing the benefits, which is impractical or just too time-consuming for the average ecom.

    One report I saw did give an aggregated stat that for each second of page load delay, an ecom site could lose as much as 7% in conversions, which is quite significant!

    If you are an ecom with global customers it makes sense to me to optimise page load times across different territories – however that’s easier said than done!