Does Speed Improve Every Website?

There have been a number of studies and surveys conducted to understand the impact of website performance on the profitability of websites. We know that the studies have consistently discovered that faster websites deliver (among other things) lower bounce rates, increased pages viewed per session and higher rates of onsite conversion in ecommerce websites.

If the studies are all delivering the same results, will these results apply to all websites?

Sobering facts from the big guys

  • Amazon.com – Revenue increased by 1% for every 100ms page speed improvement Source; Amazon
  • Shopzilla – Reducing page load time from 6 seconds to 1.2 seconds delivers a 12% increase in revenue and a 25% increase in pages viewed Source; Shopzilla
  • Walmart – Browsers experiencing average page speeds of 1-2 seconds are more than twice as likely to convert as those at 3-4 seconds Source; Walmart
  • AOL – Visitors in the top 10% of site speed view 50% more pages Source; AOL
  • Yahoo – Traffic increased by 9% for every 400ms page speed improvement Source; Yahoo
  • Mozilla – 2.2 second page speed reduction delivers an estimated 60 million more Firefox downloads Source; Mozilla

Preparing the Metrics

Preparing and running a statistically significant true A/B test to understand the impact of web performance on your site can be a costly and difficult exercise. It is not enough to speed up the pages one day and slow them down the next as there are far too many variables a play such as;

  • Day of the week (or time of day)
  • Home or landing page content visible
  • The weather
  • Competitor activity
  • TV shows timetables
  • News events etc etc

Some of these factors you can try and hold constant to deliver as consistent experience as possible. Others however, are beyond our control.

To conduct a true A/B study, you need to structure the study such that a random group of users experience a different web speed at the same time with the same content as a control group of users. Whats more the users in each group should be kept constant while they are browsing the site so the experience of the user across the total visit falls into the A or B group.

For most website owners, the cost of setting up, running and dissecting the results from such a test will likely exceed the cost of doing something about website performance in the first place.

So, its no surprise that it is largely the big guys who have tackled measurement of the correlation of website performance against web profitability. The rest of us are left to wonder whether website performance really affects every website in the same way as claimed by the largest international enterprises.

Results from Small-Medium Enterprise and Enterprise

We have been studying website performance among our customers and not surprisingly, have found similar themes to those found by Amazon et al across all our customers, regardless of size, industry, mix of Desktop versus Mobile, etc.

Here are just a few of the results;

Section ran a study with a Section media customer;

<span style=color: #000080;>Increasing Page Speed by 10% led to

  • Bounce Rate Increased by 2% from 56.7% to 58.7%
  • Pages Viewed Per Session Fell by 13.4% from 6.13 to 5.31 pages per session

Following are metrics from a large Australian ecommerce site showing the correlation between the speed of the first page (as a proxy for the whole session) and propensity to convert on site;

This graph demonstrates the clear link between page speed and the user engagement on site. As page speed increases for this Australian ecommerce site, user engagement falls (as measured by the number of page views per session).

Similarly, the following graph shows the propensity for users to bounce as related to the speed of the first page browsed for this Section customer in the Ladies Fashion industry.

The results above for user engagement, conversion and bounce rate are consistent across all Section customers. While the rate of conversion, page views per session and bounce rates all vary, the shape of the graphs correlating the effect of page speed on the metrics is consistent.

Conclusion

Based on all of the metrics available from big or small websites, there is every indication that regardless of size or type of website, improved website performance will deliver improved business outcomes.

However, until you see these results for yourself on your own web property you can never truly be sure. We would encourage businesses to engage in website performance improvement practices which are simple to deploy and which provide an opportunity to experience the improvements without being commercially committed.