A marketer’s ability to have their company’s website’s pages load quickly has direct impact on sales and ultimately on the bottom line. Page load times can impact the overall customer experience, as well as SEO as page speed determines your ranking when users search in a browser. According to the Aberdeen Group, a 1-second delay in page-load time equals 11% fewer page views, a 16% decrease in customer satisfaction, and 7% loss in conversions. And that sounds bad for business.
We partnered with the UK’s leading glasses e-tailor Glasses Direct to study page speed and conversion behavior. Our study enabled us to build a case for the true correlation between page speed and dollar values. Our findings substantiated the Aberdeen findings, where increases in page-load time significantly impacted conversion rates, dropping by 7%.
In order to get these findings, we monitored both page performance and revenue on the Glasses Direct Site. From the monitoring, we measured the impact of average paid-load time a user experienced to their likelihood to convert on the site.
We found there were significant correlations between page-load time and the likelihood of a user to convert. Here’s what we learned:
- When page-load time increases, likely conversion rates dropped by 6.7%. In the above graph you can see conversion peaking at around 2 seconds. This confirms the findings summarized in the KISSMetrics Infographic showing 47% of consumers expect a page to load in 2 seconds or less. After this peak the rate drops by 6.7% for each additional second.
- Page-load time for non-converters, users who abandoned the page without converting, was 3-to-4 times higher than for converters. This indicates that visitors who aren’t taking any actions on your site may have abandoned because of the page delays.
We calculated lost revenue for each additional second a site takes to load. Let’s say a website’s average ticket size is $75 and conversion rate is 5%. So if it takes their pages an additional second to load then for every 400,000 unique visits each month, there would be a loss of $1.3m in revenue per year.
Or to put it in other words, you could gain this revenue by speeding up your site just by 1 second. These findings are very important to TagMan as we have developed our latest version of our Tag Management Platform which optimizes page load time for all tags Smart Tag Loading.
Glasses Direct, our partner in this study, speaks on the findings. “Site speed is the most important aspect of our business,” says Oliver Elliott, Head of Digital Marketing at Glasses Direct. “Seeing this direct impact on our site visitors’ experience and how it impacts their ability to convert to customers makes us even more obsessed about our site’s speed and the overall experience we’re delivering.”
To find out how much impact an additional second load time can have on your revenue, try the calculations for yourself using our Conversion Loss Calculator. You can use a slider to apply the Glasses Direct findings to your website stats and find out how much revenue you could gain.
Hi – the problem with this (and to a lesser extent, the recent walmart study) is that you conflate cuasality and correlation.
Your study shows that people with slower pages converted less, not that slower page loads caused people to convert less. Subtle but important.
I do agree that you make a strong case but be careful about this aspect of the data – if all these customers have crud ISPs or are using high latency connections, then the problem is not your website speed per se.
Have to agree with previous comment. While it could be a great study, I’m not sure that you can in any way prove causality based on the overview you’ve given.
From the looks you’ve simply correlated page load time to conversion rate. Sadly, if true, this lacks any scientific basis to make claims.
Apologies if the study was controlled, and if so it would be very good to see both your methodology and data.
Craig, agree on the causality question. We’re are working on that next level – it’s a complicated thing to prove outright but we like complicated problems and we should have something to share soon enough.
Quite Craig. Their site may well be slow in Bolivia. Their conversion rate may also be low in Bolivia, but this doesn’t mean that speed was the cause!
Mr Sullivan beat me to it with his point about cause and effect.
The challenge with any of these speed related tests is proving that speed was the single biggest factor that caused the rise or drop in conversion rates.
There was a famous test carried out in 2003 where they took website that converted well and slowed it down and a poor converting website and made it faster and there was barely a difference in conversion rates.
They sensibly did these as usability tests and interviewed the subjects. Despite slowing down the ‘good converter’ people perceived it as fast because people could get to the information quickly
Whilst the causality point is a good one (and must never be forgotten in any analytical study worth its salt) there are plenty of examples where speed does make a massive difference.
To my mind this Amazon example is probably one of the most effective demonstrations of how performance can affect conversion rates (referenced here, although there are probably more detailed descriptions of the study somewhere else http://www.websiteoptimization.com/speed/tweak/psychology-web-performance/).
Foviance and CA also conducted a study showing how poor performance can lead to ‘web stress’ (http://www.foviance.com/what-we-think/its-official-web-stress-is-bad-for-business-2/).
A well optimised site will also cost the site owner less (most optimisation techniques also reduce server side overheads / resource usages and bandwidth).
All in all, a win win scenario – it’s good to see these studies being conducted and published for our peer review!
Craig points out one of the issues with this methodology and the response from Paul indicates this is not lost on the research. What these type of studies do afford us is the ability to justify deeper and better research methods to validate the high level assumptions.
I’m looking forward to more detailed research that can isolate more specific cause/effect relationships. This is an interesting area to watch for sure.
But – and a big but – there is clearly a sizeable chunk in this who probably did get sucky performance. It would be useful to correlate session IDs with backend response times to see if this was client, network or server side (the slower sessions).
If you had to make a call with limited info, this would be enough to push for changes. It’s not enough to make a business case from in terms of ROI – you’d really need to A/B split or similar.
@ Craig – absolutely, you need to have the complete picture of how your platform is looking and be able to identify where your bottlenecks are.
Multivariate testing (extended A/B testing) is perfect for this type of study and I’m fairly sure that several have been conducted already. The general consensus is usually the same though, that poor performance has a direct impact on bounce rates, time on site, basket size etc.
This is a nice step toward validating the reason Google has been pushing so hard to get people to pay attention to site speed. It can substantially impact your site goals/outcomes. However, as several people pointed out, there are many factors to consider and it really depends on the site.
For some sites, the impact of improving site speed by one or two seconds may make a substantial positive impact on conversion. On other sites it may have negligible impact. For example, we have a large client with millions of monthly visitors that had key pages where load time slowed from just under ten seconds to over 12 seconds, and both bounce rate and conversion improved. Obviously, the slower page load didn’t decrease bounce rate or assist in the conversion improvement. Something else coincidentally fluctuated; traffic sources.
The reported data in this test is aggregate. Aggregate and average data lie. So it would be interesting to segment the data by visitor type, traffic source, country, language, device, connection speed, lifecycle stage and etc to see if you can get a more accurate picture of where site speed improvements really make a difference.
Would love to hear any results of such segmentation.
Pingback: Four Things Marketers Should Ask IT About Their Website’s Performance | Tag Management
Pingback: TagMan shows travel company reluctance for social plug-ins [INFOGRAPHIC] « Tnooz
Pingback: Why Aren’t Travel Sites More Socially Connected? (Infographic) | Tag Management
Pingback: Why Aren’t UK Travel Sites More Socially Connected? (Infographic) | Tag Management