Blog

Case Study: Mobile Web Performance Monitoring and Conversion Rate

September 1, 2015 by Tammy Everts

In the past, gathering data about mobile performance and business metrics has been an iffy proposition. The proportion of mobile traffic that converts to paying customers hasn’t always been significant enough to provide the basis for a meaningful study into the impact of website performance on m-commerce.

Sure, you could always make the argument that delivering a faster mobile experience is an essential ingredient in delivering a better user experience overall. After all, the average online shopper visits a site 6.2 times, using 2.6 different devices, before completing a transaction. But this argument doesn’t convince the die-hards who still believe the two myths that mobile shoppers are tolerant of slow load times, and slow mobile load times don’t have a significant impact on conversions (and ultimately revenue).

Today’s post will disprove those myths.

Our test subject

For this research, I focused on a SOASTA mPulse customer, a leading online retailer that enjoys a significant amount of mobile traffic.

I looked at 30 days worth of the site’s mobile performance data, comprising almost 4.5 million mobile user sessions. Using mPulse, I plotted those sessions on a graph that shows distribution according to load time. Then I overlaid that graph with a line graph showing the average conversion rate for each cohort of pages along that load-time distribution. Then I did the same using mPulse data for the site’s bounce rate over the same set of sessions.

Here’s what I found…

Key findings

1. In terms of conversions, the performance sweet spot was 2.4 seconds

Pages that loaded in 2.4 seconds on average enjoyed the peak mobile conversion rate (1.9%) during this 30-day span. Note that a 1.9% mobile conversion rate is pretty respectable. This number approaches being commensurate with a typical desktop conversion rate of 2-3%, which we usually see for retail sites.

graph

2. Pages that were just one second faster experienced a 27% conversion rate increase

As you can see on the graph above, the mobile conversion rate drops to 1.5% for visitors who experience average page load times of 3.3 seconds. In other words, for this site, the conversion rate was 27% higher for visitors who enjoyed a load time that was about one second faster.

graph

3. At 4.2 seconds, the average conversion rate dropped below 1%

Pages that were just 2 seconds slower experienced mobile conversion rates that were cut by more than half. Or to put a more positive spin on this finding, pages that were just 2 seconds faster more than doubled their conversion rate. This, to me, is hugely significant because it puts lie to the belief that mobile shoppers are willing to be patient with slow load times. In this case, clearly, they aren’t.

graph

4. By the 6-second mark, the mobile conversion rate begins to plateau

In the past, I’ve referred to this as the “performance poverty line” — the point at which conversions more or less bottom out. As you can see below, at 5.7 seconds the conversion rate begins to flatten out at 0.6%. There’s small movement beyond that point — dipping down to its lowest point (0.5%) at 8.1 seconds and including an interesting blip upward (0.6%) at 8.4 seconds — but for all intents and purposes, this line is flat.

graph

5. Bounce rate is affected even more dramatically than conversions

Looking at bounce rate across the same load time distribution, we see an inverted version of the conversion rate line graph, with bounce rates rising steeply as performance degrades. If anything, the impact is even more deeply felt here. Pages that loaded in 2.4 seconds experienced a 12.8% bounce rate, while those that loaded in 3.3 seconds had a bounce rate of 20%. In other words, pages that were just one second slower experienced a 56% increase in bounce rate. This is huge.

And, unlike conversion rate, bounce rate didn’t plateau but instead continued to climb — eventually hitting 58% at 9.9 seconds.

graph

What about all those fast pages with low conversion rates and high bounce rates?

Good question. Faster pages should retain and convert more visitors, right? In general, yes, but some of the speediest pages on a site are 404/error pages, hence the poorer business metrics.

Takeaway

Every site is different. A 1-second improvement for this retailer correlated to a 27% increase in conversion rate. Your mileage may vary. The best way to understand how load time correlates to conversions on your site is to look at your own real user data. The important thing to remember is that speed matters, even on mobile — and perhaps especially on mobile. Even if your pages are already relatively fast, optimizing them further can pay off.

 

Categories: