Ecommerce page speed has less impact on conversion rates than most people think. Lazy loading is fine and can improve user experience. But if you’ve seen statistics likes “an X ms drop in page speed will improve conversion rates by Y%” those numbers are just wrong. The correlation between page load speeds and conversion rates is much weaker than most people believe. It’s not that it’s completely untrue. It’s just that it’s not true at the magnitude that most people seem to think.
I once did an experiment with four different test groups where I added delays in page loading to make a Magento site load 0, 1, 2, or 3 seconds slower. It had almost no impact on conversion rates between the four groups. This was at a time when a recent Google study made the claim that a 1 second reduction in page load speed translated into a 1% improvement in conversion rate on average.
A Simple Test
If reducing page loads by 1 second would have some % improvement on conversion rates, then a 1 second increase in page load speeds should have approximately the same impact, only the opposite magnitude. It’s easy to test. You can test that with a simple Javascript tag. Before you overinvest in site speed improvements, go add an arbitrary delay in page load speed to a fraction of your website traffic and test whether it changes conversion rates. A test like this is cheap and you’ll know your answer in days.
I’ve been talking with a merchant that paid for a massive project to upgrade their site and improve page load speeds by a few hundred ms and the site truly is faster. Conversion rates didn’t change and now the merchant is wondering what they bought for all that money. This experience is all too common. I’m not saying it has no impact. The site is faster. It’s a better experience for their users. That better experience must have some impact. But it’s a subtle and small impact. The expected conversion rate impact was over-hyped. This merchant is seeing no clear financial results from spending tens of thousands of dollars to shave off a fraction of a second from page loads.
What Page Speed Is ‘Good Enough’?
I wrote this four years ago, but it’s still mostly true: Should I Improve My PageSpeed Score?
You would expect that since mobile is slower than desktop, performance increases might at least improve mobile conversion rates. But again, across a bunch of companies I’ve seen invest in website speed enhancements, there are almost no detectable improvements in mobile conversion rates. My conclusion is that being really fast doesn’t matter.
Being really slow is obviously bad. We all just need to be “good enough” to be at the level of our competitors, probably. I’ve found little evidence that being truly faster than our competitors gives us any conversion rate advantage.
Correlation Or Causation?
There have been a lot of studies that claim faster sites have higher conversion rates. But they are correlative, and don’t really show causation. The companies that perform studies like this (including giants like Google) never release their data or show us the methodology of the reports they quote. They almost always talk about the statistics as “evidence from an internal study” or something like that. These studies are clearly of poor quality.
Consider that more-profitable, higher-converting websites end up with more money to invest in improvements like speed. So they are better and faster on average. That’s the same correlation, but it seems a more likely causal explanation. Perhaps it’s not that companies that invest in speed improvements pull ahead by much, but that companies which are already more financially successful tend to invest in things like speed. Claims that speed improvements cause conversion rate improvements probably put the cart in front of the horse. Highly-converting sites simply have more to invest and unsuccessful sites don’t.
Raw Speed Is Not Your Best Investment
Faster sites are great. But you know what’s better? Focusing on the most cost-effective improvements and not go crazy with expensive speed interventions. If you go to extremes with techniques like lazy loading and you’re firing conversion tags late, it can cause some tagging to not fire, resulting in Analytics inaccuracies and a reduction in the signals flowing through to ad accounts. That can have a negative performance impact on the ads that you’re actually spending a ton of money on. Developers trying to optimize for page load speeds often want to push tracking tags to the bottom, and that comes with costs in reduced advertising performance. Some interventions to speed up ecommerce sites can actually reduce financial performance through mechanisms like this.
What’s the bottom line? Page load speeds don’t have the conversation rate impacts that we’ve been sold. We should be skeptical. There is likely little benefit from being all that much faster than your competitors. Just keep pace. You don’t need to overinvest in speed improvements. You don’t want customers to think your site is slow. But there’s likely little to no benefit to being fast. Just make sure you’re good enough. Other things probably matter far more.