Henry-Paulson-and-Ben-Bernanke-testifying-about-bounce-ratesI run an experimental site on the side currently made up of several WebP images per page (with no JPEG fallback), oftentimes extremely high res. Don’t ask what the pictures are of, that’s not important. Now, WebPs are terrific in that they do all sorts of magic including displaying the same quality images at only 60 to 70% of the file size of a similar jpeg. But they require more CPU and memory for a browser to decode, so a page with, say, 4MB of WebPs, that’s a big challenge for a mobile device (even a computer) to chug through briskly even with a blazing connection. The site works on a top shelf Android phone, no problem, but most Android phones in use are not top shelf, whereas most iPhones are. Likewise with the tablets.

In my Google Analytics I noticed that iPads and iPhones running Chrome were skating through 15% more pages per visit than the site average (8.51 vs 7.42 pages), and with a 25% shorter time on site versus the site average (5:03 vs 6:45), so they are seeing more, and they’re seeing it quicker. And their bounce rate is 30% better than the site average (25.64% vs 36.61%), meaning the average iOS user who lands on my secret site is significantly more likely to browse my site than leave immediately. Meanwhile, Androids (both the stock browser and Chrome for Android, but let’s just focus on Chrome) have an 8% worse bounce rate than the site average (so, 54% worse than my iOS visits) and click through almost 40% fewer pages per visit (4.62) than the site average, seeing 56% less content in a visit than my typical iOS visitor.

That’s nuts! My site may be very atypical but sites are steadily getting heavier, and I bet if you’ve got your own site’s data, while the contrast may not be as stark, you’ll see congruent comparisons.

The explanation for that discrepancy is obvious: Web performance matters, the iPads and iPhones in circulation perform pretty well – arguably well enough now that an iOS visit is just as valuable to your ecommerce site as a computer, the typical Android device on a site that’s at all resource-intensive (I’m not even displaying videos, just pictures) perform badly enough on those sites that it affects user experience to the point where users bail out prematurely with a lower likelihood of returning. That’s what people do when a user experience sucks, they avoid repeating it.

This means Android users are less likely to go on your site and see more ads because their phones, glasses, tablets and TV gadgets can’t handle the math, it means those visitors are less likely to buy whatever you might be selling, it means they’re less likely to donate to your typhoon fund so forth. I’ll wager the same is true on the apps they run that would likewise perform more poorly. Every millisecond matters. By the way, my Mac vs Windows statistics are, though not as dramatically, similarly more impressive (a Mac visit is more valuable in terms of engagement), and I think we can all admit that the average Mac out there has more muscle and pixels than the average PC, so that’s probably why.

Therefore, while Android made an impressive splash in terms of market share and activations per nanosecond, the availability of crappy Androids that render webpages slower and the lack of pressure on consumers to pay up for a phone with enough muscle to deliver an adequate web experience means there are more people out there with a crumby phone that they are much less likely to use on the web in a manner that produces revenue to websites, ad brokers, app developers, everyone. And people on proper phones get the same things done much quicker, meaning they can see more ad impressions on more pages of one site, hop to another site and see more ads there, maybe buy something, and again, meanwhile the average Android is chugging through the first site slowly, less pleasantly and bailing out prematurely — and, I’d strongly wager, left less likely to continue using their phone for any subsequent purpose. They might even put the cheapo phone down and talk to the other people at the dinner table about how much your site’s UX sucks.

If only iPhones and Nexus phones were sold, more people would pay up for a top shelf phone and app developers and web outfits would make more money. Crappy smartphones are relatively worthless to the economy, and each crappy phone in someone’s hand is a top shelf phone that definitely isn’t in their hand. Therefore not only is Android not boosting the global GDP, Android is detracting from it. And that my friends is, in my book, the real cost of Android (as well as those cheap “Tango” Windows Phones for “emerging markets”). I believe it would be helpful, and not just to my own little bounce rate but to the whole world (including North Korea), if OEMs and carriers made an effort to put nicer phones in consumer hands somehow.

This might even be worth government subsidization. Just throwing that out there…

Doug Simmons


  1. Thank you for this information.
    This is the first time someone has agreed with me that Stock Android Browser sucked. I bought a high end device, but it rapidly became out dated over the 39 months I used it. The browser was almost unuseable at the end.
    Just got a Nexus 5, I have not put it fully through all the tests yet, Chrome performance has not been 100% but it is fast. Almost instant over WiFi and OTA.
    The Nexus 5 is my mobile device today.
    I also just got a Nokia 1520 Windows Phone. I do not have a case yet, so it has not left the house. I also have not activated the SIM card as it is the smallest, and the N5 uses the medium. I have to get an adapter so I can switch between the 2 devices.
    I bought the 1520 for Internet Explorer alone. So far it is not as fast as the N5. But I have not played with it much at all, other than to visit my regular sites that I already knew worked. I have a lot of testing to do.
    Both devices failed to upload a photo to the WindowsPhoneCentral Forum. I had to use my older device.

  2. I used to feel that mobile Chrome’s failure to reflow text upon zooming was a defect, now I’m thinking that they’ve smartened up the browser enough to make it not necessary. The fewer taps people have to make, the better, and when there’s no zoom to offer a double tap for in certain instances like a forced-fixed-zoom site, then they can ditch the 300ms delay on tapping links and buttons otherwise necessary to distinguish user attempts to zoom from clicking a link.

    Also, because mobile Chrome is gaining popularity on two platforms and being shipped as the only Android browser (at least on higher-end phones), websites, as they had to do with iPhones and no Flash, may have to make sure they’re not doing anything to trip up mobile Chrome.

    It was surprising to see that, at least based just on how many pages they clicked through, mobile Chrome on iOS visitors were getting the same site experience as desktop users.

    On my site, it takes the horsepower of the GS4 for Android visits to compete in pages, duration and bounce with the iPhone and iPad (all with Chrome).

    Anyway, wondering how many hints I need to drop to get my wife to get me a Nexus 5. I’d get it myself but it would cost me points with her (she’s trying to use the situation as leverage, her blessing for a semi-major purchase).

  3. Ehh site runs fine on the gs3 stock android. And I hardly ever visit the site since I get all the posts through feedly. Are RSS feeds part of your analysis? I’m curious how many RSSers there are. I stopped using my iPad mini after iOS 7 update since wev browsing is so studdery. It’s a much better experience on the stock CM browser than chrome on either platform. Also it’s been a while since I’ve commented on this site. Mobile experience seems much improved. Props!

  4. Anthony: I wasn’t talking about MD, rather some hobby site of mine that’s resource-intensive. No RSS – I disabled it (on the hobby site). Interested in user behavior on the site directly. We’re running Wptouch here, by the way, for mobile traffic, which is a rather lightweight theme.

  5. Well, one other thing I mentioned, on desktops/laptops, on my testing site I get a significantly better bounce on Chrome visits on Macs than Chrome visits on Windows. Also, while they click through 7% fewer pages on a visit (but still above the site average), Mac/Chrome visitors spend about 20% less time on my site, so they see more content in less time.

    Why do you suppose that is? I think it’s because they have better hardware.

Comments are closed.