Tips for analyzing and tuning performance for Black Friday
Waiting for apps to load is not an option, not with most online shoppers turning their backs against poor performing sites.
There’s a storm coming. Black Friday and Cyber Monday bookend the biggest online shopping weekend of the year. In 2014, consumers spent $2.4 billion online on Black Friday alone, according to Adobe data. Conversion rates rose to 3.2%, the average order value was $142, and 27% of those sales came from mobile devices. This year is expected to be even bigger.
As the big day approaches, and shoppers save up, ready to scour the web for the best deals, it’s vital that you prepare properly. A huge influx of traffic can cause problems for websites and apps. Failure to properly analyze, test, and tune, could lead to a major missed opportunity.
Shoppers don’t have patience
The expectations of the average online shopper have changed in recent years. In 2006, most shoppers were content to wait up to 4 seconds before abandoning the page, by 2010 they expected it to load within 2 seconds, according to a Gomez report. If your page took between 2 and 10 seconds to load in 2010, you could expect up to a 38% increase in page abandonment. It’s safe to assume that patience has not grown since then.
The average impact of a 1 second delay is a 7% reduction in conversion, an 11% drop in page views, and a 16% drop in customer satisfaction. It’s not a one-off problem, either. The same study found that 88% of online consumers are less likely to return to a site after a bad experience, and 78% would switch to a competitor’s site.
Apps show similar trends to websites, with most users expecting an app to load in 6 seconds or less, according to a study by CA Technologies. If they don’t, many users will abandon the app and the brand forever.
Test and Baseline
You want to streamline the experience for consumers, and to do that, you have to understand what the average experience is like and how it measures up. Review workflow history of visitors to your website, and establish the baseline of time taken to display the pages viewed. Using your own tests, traverse the system as though you are a shopper looking for specific anticipated deals. You can also review and benchmark user expectations on a competitor’s website and compare your results. This is a good way to identify problems and weaknesses in your flow. Benchmark your performance thinking about:
- How long does your home page take to load?
- What about product searches and product pages?
- How long does the checkout process take?
- How do those times compare to your major competitors?
It’s also important to test your website or app with different counts of concurrent users and repeated (but staggered) activities, as close to the production environment as possible. You can’t test with a small number (that is way off your original baseline data collected) and simply extrapolate up, you need to throttle the load until it falls over to understand your site threshold (or realistic limitations). Don’t underestimate the potential volume of traffic, many big retailers have done that in the past and lost out on revenue because of it.
Create a plan
Think about your business priorities. What is the key functionality that must work? There may be room to free up bandwidth by focusing on critical functions. If you can identify bottlenecks and break them ahead of time, your overall performance will be much better. Optimize your code, tune your database, consider caching, and resist the temptation to hack for short term performance boosts.
It may be worth looking at how you can add more capacity with additional server support, and get a smart load balancing strategy in place. Can you re-route traffic and add resources when you need to? You also need to plan what will happen in the event of a failure. How will you recover? It’s best to be prepared for the worst.
Learn from your mistakes
This is ultimately all about minimizing risk, and you should always treat problems as an opportunity to learn. If you’ve had issues with customer satisfaction related to performance in the past, make sure you drill into those and find out if they can tell you anything important. Analyze what went well and what could have gone better last year. If you don’t have data to draw on, make sure you get metrics in place this year, and at the very least you’ll have something to work with for next year.
While it’s important to analyze and tune ahead of major shopping events like Black Friday, and you may assign extra resources temporarily for that purpose, the best approach is to constantly measure your performance and work on ways to improve it year round.