Date: 16th August 2016
Experience shows us that effective load testing can lead to significant improvements in web performance. But because what constitutes a successful result is so specific to your business and situation, it’s important to nail realistic expectations before load testing begins.
Defining a successful outcome doesn’t have to be overly complicated, it just needs to contain some specific performance-related goals. These outcomes may be quantitative or may be focused on enhancing user experience. Whatever your criteria, they should be firmly defined before load testing begins.
Define the outcomes that will deliver actionable results
Numbers can be useful metrics but they can be misleading, too. For instance, while one of your goals may be to support ‘10 000’ users at any given time, you’d probably agree that delivering a satisfactory user experience is at least as – if not more – important a consideration. A site managing ‘10 000’ concurrent visitors performing a task means nothing at all if they are all receiving a poor experience.
Think about the baseline experience you want users to enjoy, even during peak periods. Or, to put it another way, what’s the maximum you believe users will tolerate in a high load situation?
Analytics from previous years can be used to better understand how and when traffic arrives to create an accurate load test model and ensures realistic customer journeys will be tested.
Here are a few criteria that are worth considering:
1. Experience based on sessions
It may be basic but sessions or user journeys per minute or hour is one of the key metrics to consider when assessing web performance and is easy to compare from test to test. We test to see how many good user experiences there are in a given time for user journeys under different load levels. It’s a good indicator of optimal capacity because as traffic builds and your site slows down, the number of sessions per minute/hour will start falling.
2. Page delivery times
This can be a particularly handy metric when investigating how long it takes for a customer to complete a transaction. The simplest way of targeting this is to establish an average page delivery time across the site, perhaps with individual targets set for specific pages or journeys. Alternatively – and more accurately – delivery times can be set on a page-by-page basis which involves a closer inspection of user experience under load.
3. Page delivery rates
If you measure how many of a given type of page can be delivered to users over the course of a minute, you’ll have another useful indicator of site performance. A pages-per-minute target can be set at a level that reflects expected demand, so enabling your website to deliver a good user experience throughout. As with page delivery time, an overall target or individual page targets can be set.
4. Custom KPIs
Every business will have outcomes that are specific to their own goals. If you want to report on the number of orders per minute your platform can handle or pin down an acceptable error rate, we can help you create an accurate load test model that will deliver the metrics you need.
What tools can you use to help define your outcomes?
Digging about in your analytics data is a good place to start. If you already have stacks of info on things like online transactions and user numbers – perhaps even market research on user experience – you can analyse this alongside your business goals to help define web capacity targets that are aligned with your growth projections. Outsourcing load testing to SciVisum will enable our expert team to carry out this process on your behalf.
It sounds simple, but it’s always worth taking your own customer journey through your website at different times to analyse the experience from a user’s perspective. Note response times, log any errors and evaluate the journey subjectively – would you be happy with your experience as a paying customer and how well does your web performance reflect on your brand values?
Without testing your site against realistic outcomes you may believe your site is prepared for your peak traffic but in reality it may not be able to handle traffic levels and real users performing activities across your site.
Learn more about optimising your website for peak periods and discover the secret of planning for Black Friday success at the SciVisum website. And if you’d like to read about our success stories with some of the UK’s leading retailers, browse through our case studies.