2

My application is provided as a service that is embedded in other sites. I have google analytics installed on the login popup dialog which is a page of my application, which is opened from the host site (OAuth).

About a week ago, I've noticed a sharp decrease in the number of new users registrations and a jump in the bounce rate (from ~30% to ~80%).

This happened without any change in the application. I looked into technical parameters like page load time and error rates, but could not see any change in there.

Any ideas what can cause this behavior?

4
  • How about traffic, did it grow? If you increase traffic but not watching quality, this can happen. More users, but less % interested in your product. (just a guess as we dont have much information available)
    – milo5b
    Commented Dec 2, 2012 at 17:09
  • No. Traffic is roughly the same. If there's any additional data I can provide to help understand this please let me know. I'll gladly provide it. My main problem is that every parameter I can think of stayed the same, so I guess there's some parameter I'm missing...
    – davidrac
    Commented Dec 2, 2012 at 17:40
  • @davidrac How is your script embedded? Is it a javascript snippet or an iframe? Are you saying the login box which opens on a members website opens a page or content loaded from your main application site? Have you checked the source of this popup to make sure duplicate GA snippets aren't loading?
    – Anagio
    Commented Dec 15, 2012 at 12:20
  • It it a javascript snippet. The login box opens a page from my main application page. There is no duplicated GA snippet in there. Even if there were, how would that explain the sudden change as I didn't change any of the code (including the snippet)?
    – davidrac
    Commented Dec 15, 2012 at 14:49

5 Answers 5

1

Look at Content > Site Content > All Pages and then choose a secondary dimension of Network Domain. Apply an advanced filter, including bounce rate > 75%. If you try different filters to include specific pages one by one. Look at / and other high traffic pages, you may find something is scraping your page(s) a lot (and activating the JS) from the same domain and bouncing. Most bots are excluded because they don't activate JS, and therefore aren't polluting the data.

Bounces are hits that are "1 hit and quit", not coming from any referrer and not appearing to go to another page in 30 minutes. This behavior could also be someone visiting a lone bookmarked page, but a high bounce rate for one domain as a whole may indicate they are scraping or hitting your page to get something. If that same domain with the high bounce rate has a very low average time on site (less than 10 seconds), it is likely something automated is hitting your site.

If you look at the last month of traffic, you may have some higher pageview & high bounce-rate traffic, but you will likely notice some traffic that all seem to hit you exactly 52 or 30 times each. Those are not likely human visits. With such a dramatic change, you might find one or more domains hitting your site a lot, with a bounce rate of 95-100%.

0

If traffic is roughly the same and your bounce rate has shot up then I would strongly suspect it is due to the fact that your users are not being tracked correctly from page to page (or subdomain to subdomain / site to site etc).

Basically when the user is landing on your page if the user session is dropped from the previous page then analytics can't tie the full user path together and it appears like a high bounce rate. As far as analytics can figure out the user has bounced away as it has lost the user session tracking.

Have you looked at the number of new users visiting the site? Has this increased significantly? This can be a useful check particularly if the traffic has remained more or less constant.

To fully debug I would recommend the Google Analytics Debugger plugin for Chrome which will output the full Analytics requests to the debug console.

3
  • I changed nothing, including analytics. What I'm tracking is the user registration flow. It's not only the bounce rate that shot up but also the number of new user registrations that went down, so I suspect there's more to it than analytics.
    – davidrac
    Commented Dec 14, 2012 at 5:48
  • Hi David, we would need some more detail so in order to help if you are confident that it isn't Analytics or a just a counting issue. I see below though that you are unable to pass on the URL so I'm not sure how else we can assist.
    – joesk
    Commented Dec 19, 2012 at 17:44
  • Sure. Just let me know what you need to know. I'll provide any detail required.
    – davidrac
    Commented Dec 19, 2012 at 18:11
0

You should also check if for some reason your page load time increased. Every second of load time can increase bounce rate by as much as 20%. Users like fast pages.

2
  • make sure your javascripts are running async. Use something liek pingdom tools or a yslow test Commented Dec 14, 2012 at 0:30
  • I changed nothing on the site when this happened. I checked the page load time and it is similar to what it was before.
    – davidrac
    Commented Dec 14, 2012 at 5:50
0

You could also check for double entries of the Google Analytics code. When this happens, GA tracks those pages as a page with zero bounce!

So if this was suddenly corrected, then that would explain the jump in bounce rate.

3
  • I haven't changed a thing in the site, including the google analytics code. I'm convinced this is not just a measurement issue, since the real number of newly registered users in my site went down as well.
    – davidrac
    Commented Dec 17, 2012 at 11:43
  • can you provide the link to ur app here? Commented Dec 17, 2012 at 12:20
  • It is not a standalone app. It is embedded in another company's site. I'm afraid I can't provide the URL.
    – davidrac
    Commented Dec 17, 2012 at 12:41
0

For me it was my sites login page. I upgraded and the GA snippet was being added to the user login page, which I suspect was getting hit by bots. This caused a spike in bounces from 2% to 50% overnight after I upgraded my site.

1
  • The person asking says that they didn't change the web application at all, so I don't think that could possibly be their problem. Commented Feb 29, 2020 at 12:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.